phd proposal artificial intelligence

PhD in Artificial Intelligence

To enter the Doctor of Philosophy in Artificial Intelligence, you must apply online through the UGA  Graduate School web page . There is an application fee, which must be paid at the time the application is submitted.

There are several items which must be included in the application:

• Standardized test scores, including the GRE. 
• 3 letters of recommendation, preferably from university faculty and/or professional supervisors. We encourage you to submit the letters to the graduate school online as you complete the application process.
• A sample of scholarly writing, in English. This can be anything you've written but should give an accurate indication of your writing abilities. The writing sample can be a term paper, research report, journal article, published paper, college paper, etc.
• A completed  Application for Graduate Assistantship , if you are interested in receiving funding. 
• A Statement of Purpose.
• A Resume or Curriculum Vitae.

Further information on program admissions is found in the AI Institute Frequently Asked Questions (FAQ) . 

International Students should also review the links on the  Information for International Students  page for additional information relevant to the application process.

Graduate School Policies

University of Georgia Graduate School policies and requirements apply in addition to (and, in cases of conflict, take precedence over) those described here. It is essential that graduate students familiarize themselves with Graduate School policies, including minimum and continuous enrollment  and other policies contained in the Graduate School Bulletin.

Students should also familiarize themselves with Graduate School Dates and Deadlines relevant to the degree.

Degree Requirements

Students of the doctoral program must complete a minimum of 40 hours of graduate coursework and 6 hours of dissertation credit (for a total of 46 credit hours), pass a comprehensive examination, and write and defend a dissertation. In addition, the University requires that all first-year graduate students enroll in a 1-credit-hour GradFirst seminar . Each of these requirements is described in greater detail below.

The degree program is offered using an in-person format, and classes are in general scheduled for full-time students. There are currently no special provisions for part-time, online, or off-campus students. Students are expected to attend all meetings of classes for which they are registered.

Program of Study

The Program of Study must include a minimum of 40 hours of graduate course work and a minimum of 6 hours of dissertation credit. Of the 40 hours of graduate course work, at least 20 hours must be 8000-level or 9000-level hours.

Required Courses

The following courses must be completed unless specifically waived for students entering the program with a master’s degree in Artificial Intelligence or a related field, or for students with substantially related graduate course work. All waived credits may be replaced by an equal number of doctoral research or doctoral dissertation credits (ARTI 9000, Doctoral Research or ARTI 9300, Doctoral Dissertation). Substitutions must be approved for a particular student by that student's Advisory Committee and by the Graduate Coordinator.

• PHIL/LING 6510  Deductive Systems (3 hours)
• CSCI 6380  Data Mining (4 hours) or CSCI 8950  Machine Learning (4 hours)
• CSCI/PHIL 6550  Artificial Intelligence (3 hours)
• ARTI 6950  Faculty Research Seminar (1 hour)
• ARTI/PHIL 6340 Ethics and Artificial Intelligence (3 hours)

Elective Courses

In addition to the required courses above, at least 6 additional courses must be taken from Groups A and Group B below, subject to the following requirements. 

• At least 2 courses must be taken from Group A, from at least 2 areas.
• At least 2 courses must be taken from Group B, from at least 2 areas.
• At least 3 courses must be taken from a single area comprising the student’s chosen area of emphasis .

Since not all courses have the same number of credit hours, Ph.D. students may need to take additional graduate courses to complete the 40 hours.

AREA 1: Artificial Intelligence Methodologies

• CSCI 6560  Evolutionary Computing (4 hours)
• CSCI 8050  Knowledge Based Systems (4 hours)
• CSCI/PHIL 8650  Logic and Logic Programming (4 hours)
• CSCI 8920  Decision Making Under Uncertainty (4 hours)
• CSCI/ENGR 8940  Computational Intelligence (4 hours)
• CSCI/ARTI 8950  Machine Learning (4 hours)

AREA 2: Machine Learning and Data Science

• CSCI 6360  Data Science II (4 hours)
• CSCI 8360  Data Science Practicum (4 hours)
• CSCI 8945  Advanced Representation Learning (4 hours)
• CSCI 8955  Advanced Data Analytics (4 hours)
• CSCI 8960  Privacy-Preserving Data Analysis (4 hours)

AREA 3: Machine Vision and Robotics

• CSCI/ARTI 6530  Introduction to Robotics (4 hours)
• CSCI 6800  Human Computer Interaction (4 hours)
• CSCI 6850  Biomedical Image Analysis (4 hours)
• CSCI 8850  Advanced Biomedical Image Analysis (4 hours)
• CSCI 8820  Computer Vision and Pattern Recognition (4 hours)
• CSCI 8530  Advanced Topics in Robotics (4 hours)
• CSCI 8535  Multi Robot Systems (4 hours)

AREA 4: Cognitive Modeling and Logic

• PHIL/LING 6300  Philosophy of Language (3 hours)
• PHIL 6310  Philosophy of Mind (3 hours)
• PHIL/LING 6520  Model Theory (3 hours)
• PHIL 8310  Seminar in Philosophy of Mind (max of 3 hours)
• PHIL 8500  Seminar in Problems of Logic (max of 3 hours)
• PHIL 8600  Seminar in Metaphysics (max of 3 hours)
• PHIL 8610  Epistemology (max of 3 hours)
• PSYC 6100  Cognitive Psychology (3 hours)
• PSYC 8240  Judgment and Decision Making (3 hours)
• CSCI 6860  Computational Neuroscience (4 hours)

AREA 5: Language and Computation

• ENGL 6885  Introduction to Humanities Computing (3 hours)
• LING 6021  Phonetics and Phonology (3 hours)
• LING 6080  Language and Complex Systems (3 hours)
• LING 6570  Natural Language Processing (3 hours)
• LING 8150  Generative Syntax (3 hours)
• LING 8580  Seminar in Computational Linguistics (3 hours)

AREA 6: Artificial Intelligence Applications

• ELEE 6280  Introduction to Robotics Engineering (3 hours)
• ENGL 6826  Style: Language, Genre, Cognition (3 hours)
• ENGL/LING 6885  Introduction to Humanities Computing (3 hours)
• FORS 8450  Advanced Forest Planning and Harvest Scheduling (3 hours)
• INFO 8000  Foundations of Informatics for Research and Practice
• MIST 7770  Business Intelligence (3 hours)
• MIST 7440  AI in Business and Society (3 hours)

Students may under special circumstances use up to 6 hours from the following list to apply towards the Electives group requirement. 

• ARTI 8800  Directed Readings in Artificial Intelligence
• ARTI 8000  Topics in Artificial Intelligence

Other courses may be substituted for those on the Electives lists, provided the subject matter of the course is sufficiently related to artificial intelligence and consistent with the educational objectives of the Ph.D. degree program. Substitutions can be made only with the permission of the student's Advisory Committee and the Graduate Coordinator.

In addition to the specific PhD program requirements, all first-year UGA graduate students must enroll in a 1 credit-hour GRSC 7001 (GradFIRST) seminar which provides foundational training in research, scholarship, and professional development. Students may enroll in a section offered by any department, but it is recommended that they enroll in a section offered by AI Faculty Fellows for AI students. More information is available at the  Graduate School website .

Core Competency

Core competency must be exhibited by each student and certified by the student’s advisory committee. This takes the form of achievement in the required courses of the curriculum. Students entering the Ph.D. program with a previous graduate degree sufficient to cover this basic knowledge will need to work with their advisory committee to certify their core competency. Students entering the Ph.D. program without sufficient graduate background to certify core competency must take at least three of the required courses, and then pursue certification with their advisory committee. A grade average of at least 3.56 (e.g., A-, A-, B+) must be achieved for three required courses (excluding ARTI 6950). Students below this average may take the fourth required course and achieve a grade average of at least 3.32 (e.g., A-, B+, B+, B).

Core competency is certified by the unanimous approval of the student's Advisory Committee as well as the approval by the Graduate Coordinator. Students are strongly encouraged to meet the core competency requirement within their first three enrolled academic semesters (excluding summer semester).  Core Competency Certification must be completed before approval of the Final Program of Study.

Comprehensive Examination

Each student of the doctoral program must pass a Ph.D. Comprehensive Examination covering the student's advanced coursework. The examination consists of a written part and an oral part. Students have at most two attempts to pass the written part. The oral part may not be attempted unless the written part has been passed.

Admission to Candidacy

The student is responsible for initiating an application for admission to candidacy once all requirements, except the dissertation prospectus and the dissertation, have been completed.

Dissertation and Dissertation Credit Hours

In addition to the coursework and comprehensive examination, every student must conduct research in artificial intelligence under the direction of an advisory committee and report the results of his or her research in a dissertation acceptable to the Graduate School. The dissertation must represent originality in research, independent thinking, scholarly ability, and technical mastery of a field of study. The dissertation must also demonstrate competent style and organization. While working on his/her dissertation, the student must enroll for a minimum of 6 credit hours of ARTI 9300 Doctoral Dissertation spread over at least 2 semesters.

Advisory Committee

Before the end of the third semester, each student admitted into the program should approach relevant faculty members and form an advisory committee. Until the committee is formed, the student will be advised by the graduate coordinator. The committee consists of a major professor and two other faculty members, as follows:

• The major professor and at least one other member must be full members of the Graduate Program Faculty.
• The major professor and at least one other member must be Institute for Artificial Intelligence Faculty Fellows.

Deviations from the 3-member advisory committee structure, including having more members, are in some cases permitted but must conform to Graduate School policies. 

The major professor and advisory committee shall guide the student in planning the dissertation.  The committee shall agree upon, document, and communicate expectations for the dissertation. These expectations may include publication or submission requirements, but, should not exceed reasonable expectations for the given research domain. During the planning stage, the student will prepare a dissertation prospectus in the form of a detailed written dissertation proposal. It should clearly define the problem to be addressed, critique the current state-of-the-art, and explain the contributions to research expected by the dissertation work. When the major professor certifies that the dissertation prospectus is satisfactory, it must be formally considered by the advisory committee in a meeting with the student. This formal consideration may not take the place of the comprehensive oral examination.

Approval of the dissertation prospectus signifies that members of the advisory committee believe that it proposes a satisfactory research study. Approval of the prospectus requires the agreement of the advisory committee with no more than one dissenting vote as evidenced by their signing an appropriate form to be filed with the graduate coordinator’s office.  

Graduation Requirements - Forms and Timeline

Before the end of the third semester in residence, a student must begin submitting to the Graduate School, through the graduate coordinator, the following forms: (i) a Preliminary Program of Study Form and (ii) an Advisory Committee Form. The Program of Study Form indicates how and when degree requirements will be met and must be formulated in consultation with the student's major professor. An Application for Graduation Form must also be submitted directly to the Graduate School. Forms and Timing must be submitted as follows:

• Advisory Committee Form (G130)—end of third semester
• Core Competency Form (Internal to IAI)—beginning of fourth semester
• Preliminary Doctoral Program of Study Form—Fourth semester
• Final Program of Study Form (G138)—before Comprehensive Examination
• Application for Admission to Candidacy (G162)—after Comprehensive Examination
• Application for Graduation Form (on Athena)—beginning of last semester
• Approval Form for Doctoral Dissertation (G164)—last semester
• ETD Submission Approval Form (G129)—last semester

Students should frequently check the Graduate School Dates and Deadlines webpage to ensure that all necessary forms are completed in a timely manner.

Student Handbook

Additional information on degree requirements and AI Institute policies can be found in the AI Student Handbook .

For information regarding the graduate programs in IAI, please contact: 

Evette Dunbar [email protected] Boyd GSRC, Room 516 706-542-0358

We appreciate your financial support. Your gift is important to us and helps support critical opportunities for students and faculty alike, including lectures, travel support, and any number of educational events that augment the classroom experience.  Click here to learn more about giving .

Every dollar given has a direct impact upon our students and faculty.

• Aviation and Astronautical Sciences
• Computer Science, Artificial Intelligence and Data Science
• Construction and Facilities
• Critical Infrastructure
• Cyber & Information Security
• Cyberpsychology
• Engineering
• Engineering Technologies
• Intelligence and Global Security Studies
• Management of Technology
• Occupational Safety and Health
• Uncrewed Systems
• Doctoral Degrees
• Master's Degrees
• Bachelor's Degrees
• Online Programs
• Associate Degrees
• Certificates
• Minor Degrees
• STEM Events
• Webinars and Podcasts
• Master's
• Undergraduate
• Transfer Students
• Military and Veterans
• International Students
• Admissions Counselor
• Capitol Connections
• Accepted Students
• Project Lead the Way
• Builder Culture
• Campus Life
• Clubs and Organizations
• Centers and Labs
• Online Classes
• The Capitol Commitment
• Top Employers
• Co-ops and Internships
• Professional Education
• Find a Mentor
• Career Services
• Capitol Online Job Board
• Recruiters and Employers
• Why Capitol Tech
• At a Glance
• Mission, Vision and Goals
• Diversity, Equity and Inclusion
• Washington, D.C.
• Capitol History
• Capitol Partners
• News and Events
• Visitors/Campus
• Accreditation
• Recognitions & Awards
• Current Students
• Faculty & Staff
• Alumni & Giving
• News & Events
• Capitology Blog
• Maps / Directions

• Degrees and Programs

Doctor of Philosophy (PhD) in Artificial Intelligence

• Request Information

Earn a doctorate degree in Artificial Intelligence, help lead innovation in a growing industry

The PhD in Artificial Intelligence is centered upon how computers operate to match the human decision making process in the brain. Your research will be led by AI experts with both research and industrial expertise. This emerging subject is starting to attract attention on the wider issues as the IOT and other advanced computer systems work in our lives.

This is a research based doctorate PhD degree where you will be assigned an academic supervisor almost immediately to guide you through your program and is based on mostly independent study through the entire program. It typically takes a minimum of two years but typically three years to complete if a student works closely with their assigned academic advisor. Under the guidance of your academic supervisor, you will conduct unique research in your chosen field before submitting a Thesis.

As your PhD progresses, you move through a series of progression points and review stages by your academic supervisor. This ensures that you are engaged in a process of research that will lead to the production of a high-quality Thesis and/or publications and that you are on track to complete this in the time available. Following submission of your PhD Thesis or accepted three academic journal articles, you have an oral presentation assessed by an external expert in your field.

Why Capitol?

Learn around your busy schedule

Program is 100% online, with no on-campus classes or residencies required, allowing you the flexibility needed to balance your studies and career.

Proven academic excellence

Study at a university that specializes in industry-focused education in technology fields, with a faculty that includes many industrial and academic experts.

Expert guidance in doctoral research

Capitol’s doctoral programs are supervised by faculty with extensive experience in chairing doctoral dissertations and mentoring students as they launch their academic careers. You’ll receive the guidance you need to successfully complete your doctoral research project and build credentials in the field. 

Key Faculty

Dissertation Chair

Degree Details

This program may be completed with a minimum of 60 credit hours, but may require additional credit hours, depending on the time required to complete the dissertation/publication research. Students who are not prepared to defend after completion of the 60 credits will be required to enroll in RSC-899, a one-credit, eight-week continuation course. Students are required to be continuously enrolled/registered in the RSC-899 course until they successfully complete their dissertation defense/exegesis.

The student will produce, present, and defend a doctoral dissertation after receiving the required approvals from the student’s Committee and the PhD Review Boards.

Prior Achieved Credits May Be Accepted

Doctor of Philosophy - 60 credits

Program Objectives:

• Students will integrate and synthesize alternate, divergent, or contradictory perspectives or ideas fully within the field of Artificial Intelligence.
• Students will demonstrate advance knowledge and competencies in Artificial Intelligence.
• Students will analyze existing theories to draw data-supported consultations in Artificial Intelligence.
• Students will analyze theories, tools, and frameworks used in Artificial Intelligence.
• Students will execute a plan to complete a significant piece of scholarly work in Artificial Intelligence.
• Students will evaluate the legal, social, economic, environmental, and ethical impact of actions within Artificial Intelligence and demonstrate advance skill in integrating the results in to the leadership decision-making process.

Learning Outcomes:

Upon graduation, graduates will:

• integrate the theoretical basis and practical applications on Artificial Intelligence in to their professional work;
• demonstrate the highest mastery of Artificial Intelligence;
• evaluate complex problems, synthesize divergent/alternative/contradictory perspectives and ideas fully, and develop advanced solutions to Artificial Intelligence challenges; and
• contribute to the body of knowledge in the study of Artificial Intelligence.

Tuition & Fees

Tuition rates are subject to change.

The following rates are in effect for the 2024-2025 academic year, beginning in Fall 2024 and continuing through Summer 2025:

• The application fee is $100
• The per-credit charge for doctorate courses is $950. This is the same for in-state and out-of-state students.
• Retired military receive a $50 per credit hour tuition discount
• Active duty military receive a $100 per credit hour tuition discount for doctorate level coursework.
• Information technology fee $40 per credit hour.
• High School and Community College full-time faculty and full-time staff receive a 20% discount on tuition for doctoral programs.

Find additional information for 2024-2025 doctorate tuition and fees.

When I was comparing multiple universities, Capitol Tech's admissions staff was responsive, helpful, and caring. This was valuable my deciding factor in choosing Capitol Tech.

-Dr. Jason Collins-Baker PhD in Artificial Intelligence

I chose Capitol for its great reputation in the field I am interested in, as well as its flexibility offering a fully Online program that works very well with my work and family commitments. Another reason was its European PhD program, with which I am familiar since I grew up and studied in Europe (Spain), although I currently live in the United States (California).

-Hector Garcia Villa PhD in Artificial Intelligence

Need more info, or ready to apply?

Study at Cambridge

About the university, research at cambridge.

• Undergraduate courses
• Events and open days
• Fees and finance
• Postgraduate courses
• How to apply
• Postgraduate events
• Fees and funding
• International students
• Continuing education
• Executive and professional education
• Courses in education
• How the University and Colleges work
• Term dates and calendars
• Visiting the University
• Annual reports
• Equality and diversity
• A global university
• Public engagement
• Give to Cambridge
• For Cambridge students
• For our researchers
• Business and enterprise
• Colleges & departments
• Email & phone search
• Museums & collections
• Course Directory

PhD in Human-Inspired Artificial Intelligence

Postgraduate Study

• Why Cambridge overview
• Chat with our students
• Cambridge explained overview
• The supervision system
• Student life overview
• In and around Cambridge
• Leisure activities
• Student union
• Music awards
• Student support overview
• Mental health and wellbeing
• Disabled students
• Language tuition
• Skills training
• Support for refugees
• Courses overview
• Department directory
• Qualification types
• Funded studentships
• Part-time study
• Research degrees
• Visiting students
• Finance overview
• Fees overview
• What is my fee status?
• Part-time fees
• Application fee
• Living costs
• Funding overview
• Applying for University funding
• Doctoral training programmes
• External funding and loans
• Colleges overview
• College listing overview
• Accommodation
• Applying overview
• Application deadlines
• Entry requirements
• International qualifications
• English language requirements
• Find a supervisor
• Widening access and participation
• Supporting documents overview
• Writing a research proposal
• Preparing a personal statement
• Application fee overview
• Application fee waiver
• International applications
• Disabled applicants
• AI and postgraduate applications
• Admissions fraud
• How we assess your application
• Outcome of your application overview
• Appeals and complaints
• Your offer overview
• Visas and immigration
• Declaring criminal convictions
• Defer your application
• Prepare to arrive
• Track your application
• International overview
• International events
• International student views overview
• Akhila’s story
• Alex’s story
• Huijie’s story
• Kelsey’s story
• Nilesh’s story
• Get in touch!
• Events overview
• Upcoming events
• Postgraduate Open Days overview
• Discover Cambridge webinars
• Virtual tour
• Research Internships
• How we use participant data
• Postgraduate Newsletter

Primary tabs

• Overview (active tab)
• Requirements
• How To Apply

This exciting PhD in Human-Inspired Artificial Intelligence will train the next generation of AI researchers, technologists, and leaders in the development of human-centred, human-compatible, responsible and socially and globally beneficial AI technologies. The course offers research training in areas such as fundamental human-level AI, social and interactive AI, cognitive AI, creative AI, health and global AI, and responsible AI. Students will be educated in an interdisciplinary environment where they can get access to expertise not only in the technical but also human, ethical, applied and industrial aspects of AI.

This programme is distinct from other PhD programmes in that it takes a strongly interdisciplinary and cross-disciplinary approach to technical AI. It will be based at the Centre for Human-Inspired Artificial Intelligence (CHIA) within the Institute for Technology and Humanity (ITH) where PhD students will have access to both a large community of scholars and students tackling similar questions and to the active research events programme that constitutes a key part of CHIA’s work. The course addresses the broader need for experts equipped to develop more responsible and human-centred AI as academia, industry, government and non-profit sectors increasingly recruit AI specialists and is a logical next step for students moving through AI-related master’s programmes and wishing to specialise in human-inspired AI. The interdisciplinary nature of human-inspired AI means that the programme will involve working closely also with other units of the University, including co-supervision arrangements, access to research seminars, and access to facilities.

The PhD in Human-Inspired AI aims to equip students with the skills and knowledge to contribute critically and constructively to research in human-inspired AI. It introduces students from diverse backgrounds to research skills and specialist knowledge from a range of academic disciplines and provides them with the opportunity to carry out focused research under close supervision by domain experts at the University.

The programme will train the next generation of researchers and leaders in AI by

• providing them with educational infrastructure and interdisciplinary research environment and world-leading training in human-inspired AI,
• providing them with the critical tools to engage with the forefront of academic knowledge, methods and applications in this area,
• developing the advanced skills and abilities to identify, approach and address practical interdisciplinary research challenges,
• supporting students to develop a broad and deep understanding of the technical, ethical, applied and human aspects of AI, 
• developing the ability and initiative to identify, address and approach relevant and complex challenges across sectors and society.

The course will benefit  

• students wanting to engage with human-inspired AI by enabling them to hone critical, methodological and technical skills, develop new approaches and test them out, and specialise,
• students locating themselves in other home disciplines who wish to develop advanced projects including CHIAs approaches and orientations, 
• students entering into or returning to careers in academia, tech industry, and other sectors by giving them the advanced skills, critical perspectives, and methodological insights to pursue these pathways.

Learning Outcomes

Knowledge and Understanding

By the end of the PhD programme our graduates will demonstrate:

• The ability to create and interpret new knowledge, through original research or other advanced scholarship of a quality to satisfy peer review, extend the forefront of the discipline, and merit publication.
• The general ability to conceptualise, design and implement a project for the generation of new knowledge, applications or understanding at the forefront of human-inspired AI, and to adjust the project design in the light of unforeseen problems.
• A detailed understanding of applicable techniques for cross-disciplinary research and advanced academic enquiry in the field of human-inspired AI
• The ability to make informed judgements on complex issues in human inspired AI, often in the absence of complete data.
• A critical perspective on the governance and ethical challenges that arise from applications of human-inspired AI and how these sit within and interact with wider society. 
• A systematic acquisition and understanding of a substantial body of knowledge in relation to the history, methods, and applications of human-inspired AI.

Skills and other attributes

Graduates of the course will be able to:

• Continue to undertake pure and/or applied research and development at an advanced level, contributing substantially to the development of new techniques, ideas or approaches.
• Communicate their ideas and conclusions clearly and effectively to specialist and non-specialist audiences.
• Contribute constructively within national, international and cross-disciplinary environments.
• Transfer skills and qualities acquired during the programme to successfully engage in employment requiring the exercise of personal responsibility and largely autonomous initiative in complex and unpredictable situations, in professional or equivalent environments.

Employability

Students of the programme will graduate with a formal qualification in the rapidly expanding area of AI. The emphasis is on human-inspired AI. The combination of specialist, technical expertise in AI and cross-disciplinary approaches involving a wide range of human-centric disciplines means that our doctoral graduates will be uniquely qualified in the sector. The PhD will, therefore, put them in a strong position to pursue careers in a variety of academic and non-academic settings, for example organisations and consultancies in diverse sectors such as tech, health, environment, education, journalism, civil service among others.

For those intending to continue into an academic career, the course will equip them with the skills, experience and qualification for applying for a postdoctoral research position.

For Cambridge students applying to continue from the MPhil to a PhD, students must achieve a pass in the MPhil by Thesis or an overall distinction in the MPhil by Advanced Study.

All applications are judged on their own merits, and students must demonstrate their suitability to undertake doctoral-level research.

The Centre for Human-Inspired Artificial Intelligence (CHIA) will hold an online webinar 9:00-9:45am on 4 November 2024.  Please see the  CHIA website  for information on how to register for this event. 

The Cambridge University Postgraduate Virtual Open Day usually takes place at the beginning of November.  It's a great opportunity to ask questions to admissions staff and academics, explore the Colleges virtually, and to find out more about courses, the application process and funding opportunities. Visit the  Postgraduate Open Day  page for more details.

See further the  Postgraduate Admissions Events  pages for other events relating to Postgraduate study, including study fairs, visits and international events.

Key Information

3-4 years full-time, 4-7 years part-time, study mode : research, doctor of philosophy, institute for technology and humanity, course - related enquiries, application - related enquiries, course on department website, dates and deadlines:, michaelmas 2025.

Some courses can close early. See the Deadlines page for guidance on when to apply.

Funding Deadlines

These deadlines apply to applications for courses starting in Michaelmas 2025, Lent 2026 and Easter 2026.

Similar Courses

• Human-Inspired Artificial Intelligence MPhil
• Global Risk and Resilience MPhil
• Future Infrastructure and Built Environment EPSRC CDT PhD
• Conservation Leadership MPhil
• Chemistry MPhil

Postgraduate Admissions Office

• Admissions statistics
• Start an application
• Applicant Self-Service

At a glance

• Bringing a family
• Current Postgraduates
• Cambridge Students' Union (SU)

University Policy and Guidelines

Privacy Policy

Information compliance

Equality and Diversity

Terms of Study

About this site

About our website

Privacy policy

© 2024 University of Cambridge

• Contact the University
• Accessibility
• Freedom of information
• Privacy policy and cookies
• Statement on Modern Slavery
• Terms and conditions
• University A-Z
• Undergraduate
• Postgraduate
• Research news
• About research at Cambridge
• Spotlight on...

Alternatively, use our A–Z index

Attend an open day

Discover more about postgraduate research

PhD AI in Decision Making for Complex Systems CDT / Overview

Year of entry: 2025

• View full page

The standard academic entry requirement for this PhD is an upper second-class (2:1) honours degree in a discipline directly relevant to the PhD (or international equivalent) OR any upper-second class (2:1) honours degree and a Master’s degree at merit in a discipline directly relevant to the PhD (or international equivalent).

For more information about applications, please visit the CDT in AI for Decision Making in Complex Systems website .

Full entry requirements

Please visit the CDT in AI for Decision Making in Complex Systems website for more information about applications.

Programme options

Programme overview.

Please note: We are only accepting applications for PhD in Artificial Intelligence through the Centre for Doctoral Training (CDT) in AI for Decision Making in Complex Systems.

The Centre for Doctoral Training (CDT) in AI for Decision Making in Complex Systems is a 4-year programme that will educate the next generation of AI researchers to develop and deploy new machine learning models that can efficiently cope with uncertainty in complex systems.

Bringing together researchers in machine learning from the universities of Manchester and Cambridge, the CDT will be grounded in the research areas of physics and astronomy, engineering, biology, and material science, as well as a cross-cutting theme of using AI to increase business productivity, ultimately applying the research to real-world scenarios.

For more information, visit the CDT in AI for Decision Making in Complex Systems website .

Visit our Events and Opportunities page to find out more about upcoming open days and webinars.

For entry in the academic year beginning September 2025, the tuition fees are as follows:

• PhD (full-time) UK students (per annum): Band A - TBC; Band B - £7,400; Band C - £10,500; Band D - £15,200; Band E - £25,700 International, including EU, students (per annum): Band A - £29,400; Band B - £31,500; Band C - £37,300; Band D - £45,200; Band E - £59,900

Further information for EU students can be found on our dedicated EU page.

Contact details

Programmes in related subject areas.

Use the links below to view lists of programmes in related subject areas.

• Computer Science

Regulated by the Office for Students

The University of Manchester is regulated by the Office for Students (OfS). The OfS aims to help students succeed in Higher Education by ensuring they receive excellent information and guidance, get high quality education that prepares them for the future and by protecting their interests. More information can be found at the OfS website .

You can find regulations and policies relating to student life at The University of Manchester, including our Degree Regulations and Complaints Procedure, on our regulations website .

Support PSL

Artificial Intelligence for the Sciences (AI4theSciences) doctoral program

Artificial Intelligence for the Sciences (AI4theSciences) is a doctoral program run by Université PSL. 26 PhD contracts at the interfaces of artificial intelligence or big data processing are offered. AI4theSciences is supported and jointly funded by the Horizon 2020-Marie Skłodowska-Curie Actions-COFUND European program.

Artificial Intelligence for the Sciences (AI4theSciences) participates in the development of an interdisciplinary research community composed of all PSL laboratories at the cutting edge of the use of artificial intelligence techniques in their own disciplines.

As a founding member of 3IA PR[A]IRIE and the ELLIS Paris unit, PSL is recognised at the highest international level for its work in mathematics and computer science, at the core of AI and Machine Learning. It is also a privileged place for the development of these technologies in all scientific disciplines. The AI4theSciences doctoral programme is at the heart of the Data Science program , which organises numerous multidisciplinary training activities, bringing together specialists in AI and applied disciplines.

The AI4theSciences doctoral program

Launched in October 2020, AI4theSciences is a six-year project with three calls, in 2021, 2022 and 2023. The funding of 26 doctoral contracts at the interfaces of artificial intelligence or massive data processing is assured over these two cohorts. AI4theSciences is supported and co-funded by the European Horizon 2020 - Marie Skłodowska-Curie Actions-COFUND programme.

Beyond the activities of their discipline's doctoral school, the laureates of the programme are trained in AI and Machine Learning techniques, popularisation writing, Open Science , acquiring non-academic transversal skills, etc. during two dedicated pre-entry weeks, weekly seminars and conferences organised throughout the program.

Each student benefits from double supervision: a thesis director, a PSL researcher specialising in his or her discipline, and a co-supervisor specialising in AI or massive data techniques - the latter may come from a laboratory outside of PSL or from a private partner located in France or Europe. These are open to all PSL disciplines (physics, chemistry, history, economics, etc.), as long as they involve artificial intelligence or massive data processing techniques.

Third call Admissions  

Results of the 2020 call for proposals: dissertation projects selected in the first cohort.

PhD project 1 : "3DMorphEmbryo - AI-assisted reconstruction of 3D human embryo morphology from 2D medical images to improve the prediction of its development potential" (Collège de France). Alessandro Pasqui supervised by Hervé Turlier (CIRB) and Bogdan Stanciulescu (CAOR - Centre de Robotique).

PhD project 2 : "Advanced methods for enhancing interpretability of AI tools with application to the energy sector" (Mines Paris - PSL). Konstantinos Parginos supervised by Georges Kariniotakis (Centre PERSEE Mines Paris - PSL) and Riccando Bessa (Center of Power and Energy Systems at INESC TEC).

PhD project 3 : "Dark energy studies with the Vera Rubi Observatory LSST & Euclid - Developing a combined cosmic shear analysis with Bayesian neural networks" (Observatoire de Paris - PSL). Biswajit Biswas supervised by Eric Aubourg (Laboratoire APC) and Junpeng Lao (Google Switzerland).

PhD project 4 : "The politics of coding" (Ecole Normale Supérieure - PSL). Daniele Cavalli supervised by J. Peter Burgess (République des Savoirs) and Jean-Gabriel Ganascia (LIP6, Sorbonne Université).

PhD project 5 : "Physically Informed Machine Learning for controlling unruptured intracranial aneurysms" (Mines Paris - PSL). Pablo Jeken Rico supervised by Elie Hachem (CEMEF) and Bruno Figliuzzi (Centre de Morphologie Mathématique).

PhD project 6 : "Towards neuromorphic computing on quantum many-body architectures" (ESPCI - PSL). Melissa Alzate supervised by Lionel Aigouy (Laboratoire de Physique et d'Etudes des Matériaux) and Alexandre Zimmers (Laboratoire de Physique et d'Etudes des Matériaux).

PhD project 7 : "Data-driven Enzyme Evolution" (ESPCI - PSL). Mats Van Tongeren supervised by Yannick Rondelez (Gulliver Lab) and Olivier Rivoire (CIRB).

PhD project 8 : "Machine learning for origin of life in the RNA world" (ESPCI - PSL). Francesco Calvanese supervised by Philippe Nghe (UMR Chimie Biologie Innovation) and martin Weigt (LCQB, Sorbonne Université).

PhD project 9 : "Impact of human cognitive traits on finacial market formation" (Ecole Normale Supérieure - PSL). Stefano Vrizzi supervised Boris Gutkin (LNC2) and Stefano Palminteri (LNC2).

PhD project 10 : "Language Acquisition in Brains and Algorithms: towards a systematic tracking of the evolution of semantic representations in biological and artificial neural networks" (Ecole Normale Supérieure - PSL). Linnea Evanson supervised by Yves Boubenec (Laboratoire des Systèmes Perceptifs) and Pierre Bourdillon (Hôpital Fondation Adolphe Rothschild).

PhD project 11 : "Artificial Intelligence to Decode the Genomic Replication Programme of Human Cells" (Ecole Normale Supérieure - PSL). Amir Hossein Zeraati Aliabadi supervised by Olivier Hyrien (IBENS) and Benjamin Audit (LPENSL).

PhD Project 12 : "Learning dynamics in biological and artificial neural networks" (Ecole Normale Supérieure - PSL). Pierre Orhan supervised by Yves Boubenec (Laboratoire des Systèmes Perceptifs) and Jena-Rémi King (Laboratoire des Systèmes Perceptifs, Facebook Artificial Intelligence Research).

PhD project 13 : "Unsing vocal interactions to study syntax of dolphins acoustic communication (Ecole Normale Supérieure - ENS). Chiara Semenzin supervised by German Sumbre (IBENS) and Gonzalo de Polavieja (Collective Behavior Lab, Champalimaud Foundantion).

PhD project 14 : "Processing eDNA data into relevant indicators of ecosystem health and biodiversity monitoring (EPHE - PSL). Letizia Lamperti supervised by Stéphanie Manel (CEFE) and Loic Pellissier (ETH - WSL)

Results of the 2021 call for proposals: dissertation projects selected in the second cohort

PhD project 1 : “Transfert learning in biomechanics” (Mines Paris - PSL). Matteo Bastico supervised by David Ryckelynck and Etienne Decencière.

PhD project 2 : “Physics-informed Deep Learning for the Understanding of Mesostructure Effects on the Mechanical Failure of Reinforced Polymers" (Mines Paris - PSL). Guilherme Basso Della Mea supervised by Lucien Laiarinandrasana and Petr Dokladal.

PhD project 3 : "Artificial intelligence at the service of space astrometry. A new way to explore the solar system" (Observatoire de Paris). Giulio Quaglia supervised by Valéry Lainey and Guillaume Tochon.

PhD project 4 : "Artificial Intelligence for Seismic Hazard Monitoring with InSAR" (ENS-PSL). Negin Fouladi Moghaddam supervised by Romain Jolivet and Bertrand Rouet-Leduc.  

PhD project 5 : "Whole genome sequencing and deep learning: estimating demographic and adaptive processes in range expansion scenarios" (EPHE - PSL).Alba Nieto Heredia supervised by Stefano Mona and Oscar Lao.

PhD project 6 : "AI-based damage nucleation models assessed on big 4D data and micromechanical finite element simulations" (Mines Paris - PSL). Berjo Rijnders supervised by François Willot and Thilo Morgeneyer.

PhD project 7 : "EEG classification for comatose patients" (CNRS/ENS - PSL). Kevin Reynolds supervised by David Holcman and Nathalie Kubis.

PhD project 8 :  "Quantifying Uncertainties in Physics-Informed ML" (Mines Paris - PSL). Yuke Xie supervised by Hervé Chauris and Nicolas Desassis

PhD project 9 : "Corporate Honorum. Massive Data Analysis of Careers in the Corporations" (Université Paris Dauphine - PSL). Yinglei Han supervised by Dario Colazzo and François-Xavier Dudouet. 

Results of the 2022 call for proposals: dissertation projects selected in the third cohort

PhD project 1 : “Decoding communication from brain recordings: building the next generation of brain computer interfaces “(ENS-PSL). Lucy Zhang supervised by Jean-Rémi King et Pierre Bourdillon.

PhD project 2 : “Simulating early language acquisition with data-efficient spoken language models “(ENS-PSL). Jing Liu supervised by Emmanuel Dupox et Kim Najoung.

PhD project 3 : “Emergent Behavior in Large Language Models: Social, Cultural and Psychological Dimensions “(CNRS/ENS-PSL). Noé Durandard supervised by Thierry Poibeau et Simon Hegelich.

PhD project 4 : “Linking Linguistics and Brain Dynamics with Deep Language Models “(ENS-PSL).  Pablo Jose Diego Simon supervised by Yair Lakretz et Jean-Rémi King.

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska Curie grant agreement No945304.

Governing bodies of the Artificial Intelligence for the Sciences program

The Management Team includes Thomas Walter, Scientific Coordinator of the project and Zully Rojas, the Project Manager. The Project Manager is the direct contact for any question related to the project.

The Executive Board is a joint institution made up of six researchers chosen by PSL and the Vice-President in charge of Graduate Research and Education at PSL. It is chaired by the project's Scientific Coordinator, Bruno Bouchard.

Chosen for their expertise in the research or private sector and their knowledge of issues related to AI techniques and applications, the other members of the EB are Jamal Atif, Isabelle Ryl, Judith Rousseau, Eric Brousseau and Léa Saint-Raymond.

The EB's missions are:

• Verify the scientific eligibility of projects submitted under the two calls for PhD projects
• Ensure that the submitted PhD projects meet the COFUND expectations, and that they are scientifically approved.
• Suggest for each call for doctoral projects the members of the Selection Committee
• Approve the annual budget of the COFUND
• Manage possible conflict resolution

The Selection Committee includes between 8 - 16 independent experts, internationally renowned specialists of AI and its applications, depending on the diversity of the PhD projects submitted by PSL scientists, to ensure that all relevant areas are available. The committee includes PSL academic experts, external academics, and business experts. The Selection Committee is in charge of the selection and ranking of the applicants at round 1 and 2 of the admission process. The committee will released one ranking list of the 15 in 2020 and 9 in 2021 admitted students as well as the waiting list. The list indicates which projects will or will not be set up.

Selection committee members in 2020:

• Alexandre Allauzen  (Lamsade, Université Paris Dauphine – PSL)
• Silke Biermann (CPHT, Ecole Polytechnique)
• Isabelle Bloch (LIP6, Sorbonne Université)
• François Boulanger (LPENS, ENS – PSL)
• Justine Cassel (School of Computer Science, Carnegie Mellon University)
• Jean-Michel Dalle (Agoranov director, science and technology start-ups incubator)
• Romuald Elie (Google DeepMind)
• Marie Gendrel (IBENS, ENS – PSL)
• Flora Jay (LRI, Université Paris-Saclay)
• Rémi Monasson (LPENS, ENS – PSL)
• Anne Siegel (IRISA, INRIA)

Selection committee members in 2021:

• Marc Abeille (CRITEO)
• Frédéric Lechenault (LPS, ENS -PSL)
• Thierry Morra (LPENS, ENS - PSL)
• Olga Mula (Ceremade, Université Paris Dauphine – PSL)
• Benjamin Negrevergne (Lamsade, Université Paris Dauphine – PSL)
• Alleksandra Walczak (LPENS, ENS - PSL)

Selection commitee members in 2023:

• Alexandre Allauzen (Lamsade, Université Paris Dauphine – PSL)
• Frédéric Lechenault (LPS, ENS - PSL)
• Aleksandra Walczak (LPENS, ENS - PSL)
• Madalina Olteanu (CEREMADE, Université Paris Dauphine - PSL)
• Marc Abeille (CRITERO)

The Committee on Safeguards will include the PSL HR Director, one elected doctoral student and two referents of the scientific integrity who will be independently nominated by the Vice-President for Student Life and Social and Environmental Responsibility at PSL. The Committee on Safeguards, together with the Executive Board, guarantees the fairness of the project and the respect of ethical rules. It meets when necessary and takes charge of the appeal procedures filed by candidates during the pre-selection and admission procedures.

The logic behing the Horizon 2020-MSCA-COFUND actions

Horizon 2020 is the EU Research and Innovation program. The Marie Skłodowska-Curie Actions (MSCA) - COFUND funding of Horizon 2020 aims to foster excellence in the education, mobility and career development of researchers. The funded projects have a very high level of requirement in terms of transparency in the selection, admission and career management processes of the researchers who won the call. The projects are expected to offer attractive working conditions and to promote gender equality and inclusiveness in the pursuit of excellence. These PhD projects include outreach activities and aim to inform the public about the societal stakes of research. Although they are "bottom-up", MSCA-COFUND projects must respect the "3i" rule, and be interdisciplinary, international, and intersectoral (linked to the non-academic sectors). The PhD laureates must respect the MSCA "mobility rule", i.e. not have lived more than 12 months in the project country during the last three years at the time of their application. There is no requirement on the applicants' citizenships. This is intended to promote European mobility and build the European research area (ERA).

Program partners

Air Liquide, ArcelorMittal, Fondation Adolphe de Rothschild, Total

Axa Research Fund, FAIR et Google Switzerland.

If you have any questions, please do not hesitate to contact the team at  [email protected]

• [email protected]
• +91-97 91 62 64 69

PhD Research Proposal Artificial Intelligence

One of the most important subject areas of computer science is Artificial Intelligence. It provides a wide platform for building a machine with learning capabilities . Artificial intelligence makes machines think and react similarly to humans in uncertain situations. In other words, this machine intelligence to behave artificially like human intelligence is known as artificial intelligence. This page is intended to present you useful information on PhD Research Proposal Artificial Intelligence along with the latest research areas, technologies, challenges, trends, techniques, and ideas !!!

Assume that there is a situation in which a human is performing a particular task by learning and understanding the event to solve associated problems. This human task is performed by machine artificial learning abilities are known as artificial intelligence.

For instance: a self-driving car without a driver. In this, the vehicle monitors the environment and takes effective decisions for secure destination attainment.  

What are the requirements for good research proposal writing? 

We believe that we made you are clear with the exact purpose and importance of artificial intelligence at this moment. Now, we can see about the requirements of good PhD Research Proposal Artificial Intelligence . Basically, the writing of PhD proposal needs more concern and study to create a qualified proposal. Since it is the reflection of your research activities and efforts in the form of valuable words. Here, we have given you a few important tips to prepare good proposal writing.

• Need to be adaptable to access required information and resources
• Need to be meet the expected standard and enhance interest to read
• Need to be original to create a new contribution to the handpicked research area
• Need to be related with your degree and present research areas of artificial intelligence

In general, the PhD research proposal has a standard format to write. As well, it is composed of different components such as title, abstract, introduction, literature study, methodologies, conclusion, and references. In fact, we have a native writer team to give complete assistance in perfect proposal writing. Further, we also help you in literature review writing, paper writing, and thesis writing. Here, we have given you a few important things that need to be focused on while writing PhD Research Proposal Artificial Intelligence.   

What are the Components of a Good Research Proposal? 

• Give a short and crisp title for your research proposal
• Choose a title that addresses your research problem and proposed solutions
• Provide a summary of your research work
• Act as detailed synopsis that answers why, how, and what questions of your research
• Present your selected research area and research problem(s)
• Highlight the significance of your study
• Provide sufficient hypothesis of research
• Mention the methodologies that going to be used as solutions
• Talk about the review of secondary research materials
• Address the identified research gaps in previous related studies
• Do a comparison of techniques and arguments in existing researches
• Describe the contribution and findings of the previous research
• List the merits and demerits of existing research works
• Present system architectural design
• Give a detailed explanation on used research tools and techniques/methodologies
• Speak about the need and importance of choosing those methodologies
• Explain the numerical formulas and used algorithms
• Give justification for your proposed research methodologies
• Mention in what way your research methodologies solve your research problem
• Again give an overview of your research
• Point out the objectives and importance of your research
• Encapsulate all highlights of your research in brief
• A present unique point of your study
• Overall, write nearly two paragraphs
• Provide citation of your referred research websites and books
• Implicitly these references mention your supportive hypothesis
• Narrow down your wide research sources
• Smart picking of research materials will impress the research committee

We hope that you are clear with the fundamentals of writing a good PhD research proposal artificial intelligence . Now, we can see about the three primary research terms of artificial intelligence. Since these terms are most widely used in many research areas of artificial intelligence.  As well, it is categorized into three classifications such as, 

• Exploration Areas
• Real-Time Applications

Our researchers are good at proposing modern research work in upcoming research areas for smart applications . If you are interested to know more research ideas from the following classifications, then make an online or offline connection with us.   

What are three important terminologies in Artificial Intelligence? 

• Genetic Evolutionary
• Logical Rationalism
• Molecular Biological
• Statistical Empiricism
• Neural Connectionism
• Smart System Design
• Learning Approaches
• Inference Mechanism
• Knowledge Representation
• Expert System
• Electronic Commerce
• Bioinformatics
• Intelligent Robots
• Natural Language Processing
• Information Retrieval
• Data Mining

In addition, we have also given you some significant research areas of artificial intelligence . We assure you that all these areas are recognized in current AI research topics and ideas. 

Moreover, we also support you in other important research ideas to support you in all aspects of artificial intelligence . By the by, our first and foremost task in AI research is identifying your interesting research area. Then, we provide you list of the latest research notions and phd topics in artificial intelligence .

Research Areas for PhD Research Proposal Artificial Intelligence

• Reinforcement Learning
• Supervised Learning
• Unsupervised Learning
• Dialogue Systems
• Natural Language
• Understanding
• Recognition
• Classification

Furthermore, we have given you a few important supporting AI technologies. Due to the beneficial impact of AI, it is employed and demanding in several research fields (i.e., other technologies). For your information, here we have given you only a few of them. Once you connect with us, we let you know more about up-to-date research topics of your selected technologies . Specifically, these technologies are currently successful in creating real-time AI applications for the development of a smart society.

Converging Technologies of AI 

• Internet of Things
• Big Data Analytics
• Blockchain Technology
• Lightweight Cryptography
• Cloud Computing
• Software-Defined Networking
• Fog Computing
• 6G Networks
• Industry 4.0
• UAV Communication
• Autonomous Vehicles
• Edge Networks

As a point of fact, AI is treated as the shared technology which used to solve different problems in different technologies. So, it can be recognized in many real-time applications and services. Although this field has so many developments in real-time applications, it has some technical issues that arise in the time of development and deployment . For your reference, here we have listed a few important technical issues of AI in recent research.      

Artificial Intelligence Research Problems 

• Optimized Modern Parameters
• Non-linearity from learning to compensate
• Hard-to-Model Issues
• Knowledge and Learning Representation
• Solution for Computational Infeasibility
• Computationally Understanding Solutions
• Training Policies

Already, we have seen converging technologies of artificial intelligence in an earlier section. To the continuation, now we can see about the current trends of AI. In order to identify these trends, our research team has studied the present and past 2-3 years’ research articles and magazines. Through this review, we analyzed and identified

1) Research gaps that need to address

2) Problems that need enhanced solutions than existing one

From this collection, we have listed only a few of them for your reference. Further, we are also ready to share more trends that are sought by active research scholars in the field of artificial intelligence.    

Artificial Intelligence Current Trends

• Mainly in sustainable developments, energy usage has a key player role
• Provides productive communication plans for improving energy-efficiency
• Support significant services in 6G communication
• Human-sensed data are composed with 5D services to enhance the holographic communication
• Assure high QoS, precision, deterministic in 6G communication
• Need tremendous data rates like Tb/s
• Currently, manufacturing industries are moving towards automation technologies and precision communication
• In this, 6G is assured to give ultra-low delay and ultra-high reliability
• For real-cases, the general data transmission need industrial networks for low latency jitters
• For achieving a secure environ, wireless technologies, IoT and fog-cloud computing are advancing over global sustainability and QoS
• Presently, the 6G network understands 3D communication to enhance several applications like smart transportation, smart cities, smart healthcare, etc.
• For instance – Self-driving vehicles delay < 1ms and reliability > 99.999% for fast decisions over sudden accidents

Now, we can see emerging techniques that play a major role in bringing effective research solutions for different current research problems. As a matter of fact, our developers are proficient-enough to identify the best-fitting research techniques and algorithms for any sort of research problem .

In the case of complications in solving problems, our developers analyze the degree of problem complexity and create hybrid technologies or new algorithms accordingly. Overall, we are good to tackle the problem at any level of complexity in smart ways. Also, we suggest key parameters and development tools that enhance your system performance.   

Latest Techniques in AI 

• Generally, the data are collected from different formats, mode representations and sources
• Merging all these dissimilar data in one place is a tedious task
• For the data fusion, advanced neural networks and bayesian learning is used
• For instance – CNN, RBM, and DBM
• Through sensors, collect raw data and transfer it into high-computational devices for data processing
• This may cause more power usage and high traffic load over the network
• So, it is required to design a system that minimizes load and power usage without losing vital information
• Utilize ANN and perform preprocessing
• Also, network topology and architecture are required to be chosen appropriately for add-on benefits
• Prevent interference for primary user benefits through spectrum sensing
• The significant role of the primary user is to transmit data between secondary users and the succeeding layer
• This process is executed by Cooperative Spectrum Sensing (CSS) with high power usage
• The power usage increases because of report findings and spectrum sensing with respect to a centralized location
• Similarly, Convolutional Neural Network is utilized in Deep Corporate Sensing

Additionally, we have given you some growing ideas about artificial intelligence. These ideas are selected from different trending research areas that gain more attraction from the research community. If you have your own ideas to implement an artificial intelligence project, then we support you to upgrade your idea to match the latest advancements of artificial intelligence. So, create a bond with us, to know new interesting PhD research propsoal artificial intelligence . Overall, we give assistance on not only these ideas but also beyond this list of ideas.   

Emerging Ideas on AI 

• Artificial Intelligence for Internet of Things
• Privacy-Aware AI-assisted Edge System for Trustable Services
• Fast AI Services Migration from Cloud into Edge
• Secure Data Dissemination on AI-assisted Edge Systems
• In-depth Learning Services over Edge Network
• Energy-Aware AI-assisted Edge System for Quality of Services
• Real-time AI-assisted Edge Systems with Optimized Solutions
• Edge-intensive Distributed / Collaborate / Federated Smart Services

On the whole, we are here to update you about the recent research updates of artificial intelligence in every possible area. Particularly, we help you in research problem selection, corresponding solutions selection, PhD Research Proposal Artificial Intelligence Writing, code development, paper writing, paper publication, and thesis writing. So, think smartly and hold your hands with our technical experts to shine your AI research career.

Opening Hours

• Mon-Sat 09.00 am – 6.30 pm
• Lunch Time 12.30 pm – 01.30 pm
• Break Time 04.00 pm – 04.30 pm
• 18 years service excellence
• 40+ country reach
• 36+ university mou
• 194+ college mou
• 6000+ happy customers
• 100+ employees
• 240+ writers
• 60+ developers
• 45+ researchers
• 540+ Journal tieup

Payment Options

Our Clients

Social Links

• Terms of Use

Opening Time

Closing Time

• We follow Indian time zone

Improving economic policy

Suggested keywords:

• decarbonisation
• climate change
• monetary policy
• / Publications

Catch-up with the US or prosper below the tech frontier? An EU artificial intelligence strategy

This Policy Brief explores why EU AI investment has fallen behind the US and the types of market failure that may have led to that situation

Reading time: 19 minutes

Recommended citation:

Share this page:.

Executive summary

European Union policymakers want to close the artificial intelligence innovation gap with the United States, as a way to accelerate lagging productivity growth. The EU focus is on expanding an existing supercomputer network with more AI hardware and computing infrastructure, with taxpayer support. However, this computing infrastructure is not adapted to AI modelling. The cost of catching up with leading big tech AI computing centres is already prohibitive for EU budgets, and is set to become even more so.

The hardware focus overlooks missing EU markets for complementary services that are required to set up a successful AI business: large-scale business outlets for frontier generative AI models to generate sufficient revenues to cover huge fixed model training costs, hyperscale cloud-computing infrastructure and private equity financing for AI start-ups. In the absence of (or with insufficient) complementary services markets in the EU, start-ups are forced to collaborate with US big tech firms. Injecting taxpayer subsidies to make up for these missing markets may further distort EU markets. Regulatory compliance costs, including uncertainty about the implementation arrangements for the EU Artificial Intelligence Act, add to market problems.

The EU should address a wider range of market failures in its policy initiatives. It should strive to increase productivity growth below the AI technology frontier, by facilitating investment and applications of AI-driven services produced by derived and specialised generative AI models, or AI-applications that build on top of existing generative AI models. Building these below-frontier AI applications requires far less computing capacity and less heavy investment costs. Promoting the uptake of AI application services across a wide range of industries can substantially stimulate productivity growth.

That requires a razor-sharp focus on pro-innovation guidelines, standards and implementation provisions for the EU AI Act, shortening the Act’s regulatory uncertainty horizon as much as possible, and facilitating collaborations between EU AI startups and big tech companies. Widening and deepening the EU private equity and venture capital market would also be very helpful.

1 Introduction

In the first half of 2024 alone, more than $35 billion was invested globally into artificial intelligence startups 1 Joanna Glasner, ‘AI Gobbled A Record Share Of Startup Funding This Year’, Crunchbase News, 4 September 2024, https://news.crunchbase.com/ai/record-share-startup-funding-2024-xai-an… . . The European Union attracted only 6 percent of that. The EU is doing better in AI patents and in training AI researchers, but the outputs from this tend not to stay in the EU, but rather to flow to the United States (Renda, 2024). Unsurprisingly, this situation has triggered considerable debate in EU policy circles about what can be done at EU level so the EU can catch up with the US and China on AI, in particular by developing its own AI models, fostering more AI startups, accelerating the uptake of AI-based services in the EU economy.

In this context, the European Commission in January 2024 published a package of proposals, decisions and plans to support AI startups 2 See European Commission press release of 24 January 2024, ‘Commission launches AI innovation package to support Artificial Intelligence startups and SMEs’, https://ec.europa.eu/commission/presscorner/detail/en/ip_24_383 and European Commission (2024). . This seeks to capitalise on the European High-Performance Computing (EuroHPC) network of supercomputers – very large, high-performing computers – used primarily for scientific research. The Commission proposed an amendment to the network’s governance rules to facilitate collaboration with the private sector – that amendment has since been adopted (Regulation (EU) 2024/1732). The plan is that EuroHPC should be the core of a network of ‘AI factories’ for the development by EU startups of large-scale general purpose AI models and applications.

This approach recognises that these supercomputers need to be upgraded to AI capabilities, to be financed equally by the EU and the computer-hosting EU countries 3 The nine supercomputers are hosted in different countries; see https://eurohpc-ju.europa.eu/supercomputers/our-supercomputers_en . . But the EuroHPC budget of €7 billion for 2021-2027 remains for now unchanged 4 Other initiatives, complementary to EuroHPC and including a European ‘CERN for AI’ and other moonshot AI initiatives, have been proposed. For an overview, see Renda (2024). .

The AI computing infrastructure budget could be increased very substantially if the Commission and EU countries listen to former Italian prime minister and European Central Bank governor Mario Draghi. His September 2024 report on the future of European competitiveness, produced to steer EU policy in the next five years (Draghi, 2024), attributed the EU’s weak productivity growth to insufficient investment and uptake of digital technologies, including AI.

His proposed remedies include private and public investment in EU-developed general and sectoral AI models, upgrading EuroHPC, creating an AI incubator similar to that of the CERN nuclear and particle physics laboratory, creating EU-wide large data pools for AI model training, facilitating consolidation among EU cloud providers to create hyperscale computing infrastructure and more financial resources for quantum computing. Draghi (2024) also recognised that the EuroHPC computers cannot compete with US-based hyperscale AI firms and proposed to allocate €100 billion for AI infrastructure.

All this suggests a consensus in EU policy circles that catching up on AI requires public sector involvement and subsidies. There has been less analysis, however, of why the EU AI value chain and business ecosystem have ended up falling behind the US and China in terms of AI model development 5 For more details on the global competitive landscape in AI modelling, see Martens (2024b).  and uptake in services industries, and why this should justify public sector involvement and subsidies. There is even less debate on how these problems could be addressed through structural reforms that could incentivise more private investment in EU AI industries.

This Policy Brief explores why EU AI investment has fallen behind the US and the types of market failure that may have led to that situation. We ask how the EU should position itself in the competition over AI and discuss two possible responses. Should the EU try to catch up with the US, reach the AI technology frontier and develop its own AI capacities, independent of US big tech firms? Or can the EU prosper below the AI technology frontier, in derived AI products and services markets? We also look at the geopolitical context and the risks of EU dependence on US big tech.

2 Building AI models on existing EU supercomputers?

It is clear that the EU is running behind the US in digital technology investment and uptake in general, and in AI specifically. It is less clear whether the response should be to invest taxpayer money in physical infrastructures for AI, as advocated by the Commission policy initiatives and Draghi (2024). It might be possible to resolve some market and regulatory failures in AI-related markets with public money, but many other problems cannot be resolved this way. In this section, we discuss the main considerations that should be factored in to the EU approach to AI.

2.1 Computing hardware issues

The EuroHPC network of nine supercomputers is not up to the task of delivering a state-of-the-art AI computing infrastructure for commercial use. These computers were designed for scientific research, not for training of general-purpose AI models or generative AI models like ChatGPT 6 We define Generative AI models as machine learning and neural network models that apply the ‘transformer’ architecture (Vaswani et al, 2017). . Their hardware architecture is not suitable for that purpose. They have no more than a few thousand Nvidia graphics processing units (GPUs) that play a central role in AI model training. This is a tiny capacity compared to Meta’s most advanced AI computing centre, which reportedly contains 600,000 Nvidia AI chips 7 Katie Paul, Stephen Nellis and Max A. Cherney, ‘Exclusive: Meta to deploy in-house custom chips this year to power AI drive – memo’, Reuters, 1 February 2024, https://www.reuters.com/technology/meta-deploy-in-house-custom-chips-th… . .

Hobbhahn et al (2023) explained how AI hardware differs from classic computing architectures that revolves around central processing units (CPUs). Handling the massive amounts of data in GenAI model training requires GPUs. Nvidia became successful in AI hardware because of its original specialisation in GPUs for gaming applications. Handling data traffic between many thousands of GPUs requires extensive communication bandwidth between GPUs and memory storage, though one way to reduce computational requirements can be to reduce the number of digits behind the decimal point in calculations 8 Venkataramani et al (2024) stated that: “Historically, high-performance computing has relied on high-precision 64- and 32-bit floating-point arithmetic to deliver accuracy critical for scientific computing tasks. For deep learning (DL) algorithms, however, the natural error-resilience due to the presence of statistical approximation and estimation makes high-precision computation rather unnecessary”. . AI developers are increasingly designing their own dedicated hardware, including for specific applications such as inference, meaning the making of predictions based on newly supplied data after the model has been trained.

2.2 AI infrastructure costs

Nvidia AI chips each cost more than $30,000. For Meta’s most advanced computing centre with 600,000 of these chips, this amounts to $18 billion for the dedicated AI chips alone, excluding other hardware needs. In other words, the cost of chips for a single computing centre is more than twice the current EuroHPC budget.

Moreover, technological progress in AI chips is so fast that the latest generation of AI chips will be outdated and written off in less than a year (Hobbhahn et al , 2023). Spending $162 billion per year (ie nine EuroHPC supercomputers x $18 billion/year) is simply beyond the financial resources of the EU. Even if the EuroHPC network were to be upgraded to train state-of-the-art AI models, it would still have a hard time running these models on a daily basis to respond to user queries because that requires additional investment in a different type of inference accelerator chip, such as NVIDIA’s Jetson processors, to reduce the cost of responding to user queries.

The costs of training state-of-the-art generative AI models (ie those that can produce new images, video, audio or text based on prompts) are exploding, running into hundreds of millions of euros (Martens, 2024a). Cottier et al (2024) estimated that GenAI model training costs are increasing exponentially by a factor 2.4 to 2.6 per year, or around 240 percent per year from 2016 to 2023. Extrapolating the costs of the largest frontier models now to 2030 leads to an estimated training cost for a single GenAI model of $60 billion.

New frontier GenAI models are coming out every week. Cottier et al (2024) also estimated the cost of AI computing infrastructure at ten times the cost of model training. That infrastructure can be used to train several models but the hardware amortisation rate is estimated at 140 percent per year, or 100 percent depreciation in 8.5 months. By that time, a new generation of AI computing chips will have arrived with superior performance. Infrastructure costs for GPT4 by the end of 2023 may have been as high as $800 million. Extrapolation could push that figure up to $500 billion by 2030. This is beyond the financial reach of EU public and private budgets. Even the largest US big tech firms will have a hard time financing this, and may be forced to collaborate.

2.3 Integration of AI into business models

To succeed, AI startups require not only computing infrastructure but also an important complementary asset: a business model to generate revenue that pays for these costs. Most AI start-ups have close collaboration agreements with US big tech firms, to access hyperscale computing capacity and because they can directly plug their AI models into big tech’s established business models to generate revenue. Microsoft uses AI in its business software, Meta uses it in advertising and Google in search, advertising and many other services.

AI startups can also try to launch their own business models from scratch to generate sufficient revenue to finance AI model development. But this is very hard. Even successful start-ups such as OpenAI have a hard time generating sufficient revenue, despite running a successful business model 9 See for example Vishakha Saxena, ‘OpenAI’s ‘$8.5 Billion Bills’ Report Sparks Bankruptcy Speculation’, Asia Financial, 29 July 2024, https://www.asiafinancial.com/openais-8-5-billion-bills-spark-bankruptc… . . An upgraded EuroHPC network may have the hardware capabilities but offers no commercial outlet channels. EU startups would have to move their AI models to incumbent big tech firms to generate revenue to finance the fixed training costs.

2.4 Missing markets for EU AI startups

In the absence of home-grown big tech platforms, EU AI startups have to turn to US companies with global business models that have sufficient market scale to amortise the huge fixed costs of training generative AI models. In the absence of sufficient domestic private equity and venture capital in the EU, US markets and big tech firms can provide financial resources for EU startups. EU public funds might perhaps replace private equity but cannot replace business outlets for AI models. Alternatively, EU AI startups could focus on specific AI model applications, derived from big generative AI models. This avoids very high model training costs and makes it easier to plug derived models into existing services markets, where there is demand for these specific applications.

The absence of business model considerations in the Commission’s January 2024 ‘AI factories’ initiative, and in the AI recommendations in Draghi (2024), is problematic 10 These initiatives could have learned from the poor performance of earlier initiatives, such as GAIA-X ( https://gaia-x.eu/ ), an EU-sponsored plan to create a European alternative to US-based hyper-scale cloud computing infrastructure. Take-up to date has been rather weak. See Maximilian Hille, ‘Why GAIA-X hasn’t been successful yet’, Cloudflight, 21 May 2021, https://www.cloudflight.io/en/blog/why-gaia-x-hasnt-been-successful-yet/ . . But the exclusive hardware focus of these plans is not surprising. Eckert (2024) presented an insightful historic overview of EU digital policies over the past 40 years. A recurrent pattern has been the emphasis on telecoms infrastructure and hardware in general, and the almost total absence of digital services markets and business model considerations. Draghi (2024, Part B, Figure 4) showed how the value of telecoms services has become negligible compared to digital services markets. Nevertheless, his recommendations focused on telecoms, cloud and AI hardware, and do not mention digital or AI services markets. More than anything, four decades of path dependency in EU digital policies may have contributed to an ever wider yawning gap between EU and US digital performance – which still continues today.

Draghi (2024) pointed out that the EU should do more to create its own hyperscale cloud computing infrastructure in support of generative AI model development, and reduce dependence on the US big tech firms that currently dominate the cloud services market in the EU 11 See for example Back4app, ‘Top Cloud Providers in Europe’, undated, https://blog.back4app.com/top-cloud-providers-in-europe/ . . There may be competition failures in EU cloud computing services, another important complementary input for AI. A few big tech players can leverage their positions in cloud software- and platforms-as-a-service, rather than just offering basic infrastructure-as-a-service 12 Lionel Sujay Vailshery, ‘Cloud computing market size in Europe from 2018 to 2029, by segment’, Statista, 28 June 2024, https://www.statista.com/forecasts/1235161/europe-cloud-computing-marke… . . This increases entry barriers for smaller EU cloud service providers, leaving them unable to expand their computing infrastructure, which would be suitable for AI model training (Ennis and Evans, 2024; Biglaiser et al , 2024). Throwing taxpayer money at this problem is unlikely to be a good solution, however. Draghi (2024) recommended consolidation among smaller EU cloud players. That does not solve the problem of lack of complementary software and platform services.

2.5 Derived AI model markets are very competitive

There is no indication of a market failure that would require public policy intervention, let alone taxpayer subsidies, in derived and special applications of AI models. Draghi (2024) recommended that EU AI funds could support European AI startups to develop specific industry or company application models. That market is already very competitive (Martens, 2024b). While more than a dozen new state-of-the-art GenAI models are released every month, more than a dozen derived models are released per hour 13 LifeArchitect.ai, ‘LLMs released per month (2024)’, undated, https://s10251.pcdn.co/pdf/2024-Alan-D-Thompson-LLMs-released-per-month… . . Just as app stores for mobile phones contain millions of special-purpose apps, there are now also millions of industry-, sector- or company-specific applications of the ChatGPT model in the OpenAI store.

For example, there are ChatGPT applications that help consumers with their shopping questions or financial decisions. Developers of these applications make them widely available to anyone who can use them. A derived model is created when a company uploads its own proprietary data into ChatGPT for specific marketing, logistics or industrial process applications within the company. Since they run on proprietary data, these models are of course not made widely available. 

2.6 Risk of regulatory failure

Policy intervention may create new market distortions in AI services markets. The amended EuroHPC regulation (Regulation (EU) 2024/1732) now allows collaboration between public and private computing and cloud services providers. Commercial firms can access public ly-owned computers.

This raises the question of how scarce computing capacity will be allocated between users, at what price and under what conditions. The amended EuroHPC regulation does not explain this. Will authorities use auctions for commercial applications and sell capacity at market prices? Or will there be a subsidy component in pricing, thereby opening the door to unfair competition with private providers? How will capacity be allocated between paid commercial and presumably unpaid non-commercial use, for instance for scientific projects, which the EuroHPC network was involved in from the start? More importantly for AI start-ups, what happens after the training of their AI model has been completed? Will they have guaranteed access for inference, daily running of their models? Can they easily scale up capacity when their startup rapidly expands? Computers may be provided by the public sector but they are not non-rival non-excludable public goods. They are rival and easily excludable.

The European Commission and Draghi (2024) claim that these AI policy initiatives can capitalise on EU regulation, including the EU AI Act (Regulation (EU) 2024/1689; see Box 1), the general data protection regulation (GDPR, Regulation (EU) 2016/679) and other data regulations. The claim is that these EU regulations attract investment because they give users confidence and regulatory certainty.

The available empirical evidence, however, does not support that view. There is considerable evidence that the GDPR has reduced investment in consumer-oriented online services in the EU (Demirer et al , 2024; Goldberg et al , 2023; Jia et al , 2023; Peukert et al , 2024). Consumers may be better off without some of these privacy-infringing services, though that may not be the case for all.

There is also evidence that EU regulation is limiting EU access to AI services. At the request of the Irish Data Protection Commission, Meta held back the roll-out of its most advanced AI models in the EU 14 Eliza Gkritsi, ‘Breaking: Meta halts AI rollout in Europe after ‘request’ from Irish data protection authorities’, Euractiv, 14 June 2024, https://www.euractiv.com/section/data-protection/news/breaking-meta-hal… . . European data regulators have doubts that the legitimate interest clause in the GDPR (Article 6(1)(f)) constitutes a sufficient legal basis for Meta to use publicly posted messages on its Facebook and Instagram social media platforms as inputs for AI model training. Other US AI developers, including Apple, Google and OpenAI, face similar EU uncertainty about the use of personal data for model training 15 See for example Vallari Sanzgiri, ‘OpenAI Not Releasing its Emotion-Inferring Voice Feature in the European Union’, MediaNama, 27 September 2024, https://www.medianama.com/2024/09/223-openai-voice-feature-not-availabl… , and James Morales, ‘Meta Hits Back at EU Crackdown: Requests Access To European Data for AI’, CCN.com, 20 September 2024, https://www.ccn.com/news/technology/meta-eu-crackdown-zuckerberg-reques… . . The Irish Data Protection Commission launched an enquiry into Google’s AI services 16 Data Protection Commission press release of 12 September 2024, ‘Data Protection Commission launches inquiry into Google AI model’, https://techxplore.com/news/2024-09-ireland-eu-privacy-probe-google.html . . Social media text has become an important source of AI model training data when other sources are insufficient to meet the training requirements of very large AI models, especially in relation to less-widely spoken languages, for which the available volume of human text data is limited. Regulatory uncertainty about this alternative source is holding back innovative AI services from entering the European market.

Longpre et al (2024) showed that, following the release of ChatGPT in 2023, copyright holders are making more active use of their right to exercise an opt-out for content from use for AI model training, granted to them under the EU AI Act and the EU Copyright Directive (Directive (EU) 2019/790). This has led to a 20 percent to 30 percent reduction in the availability of AI training data.

Strict enforcement of data privacy consent rules could have a similar negative effect on the availability of AI model training data. AI model training is already running into data shortages (Martens, 2024b). This could especially affect small language communities in the EU that already suffer from insufficient language training data. Moreover, the AI Act (Box 1) generates not only high compliance costs for model developers and deployers, but also considerable regulatory uncertainty regarding the specific implementation rules for copyright and privacy protection. The finalisation of the AI Act in mid-2024 was only the start of a regulatory process that will take several years to complete dozens of implementation guidelines and enforcement standards, including on copyright and data privacy.

Box 1: The EU AI Act

The Artificial Intelligence Act (Regulation (EU) 2024/1689), finalised in mid-2024, is intended to regulate AI in the EU by banning certain applications that impinge in citizens’ rights and creating a category of high-risk systems and uses, for which risk assessments and measures to offset risks will be required. Decisions taken by high-risk systems should in principle be explainable and appealable. The law also contains transparency requirements, such as labelling obligations for AI-generated images, audio or video, and obliges compliance with EU copyright rules. Parts of the law are being phased in, but it will apply in full from August 2026.

The AI Act also created an AI Office, which was established in May 2024 17 See European Commission press release of 29 May 2024, ‘Commission establishes AI Office to strengthen EU leadership in safe and trustworthy Artificial Intelligence’, https://ec.europa.eu/commission/presscorner/detail/en/ip_24_2982 . , as a monitoring, supervisory and enforcement body in relation to general purpose AI models and systems. Among its responsibilities will be development of specific implementation rules, including on AI and copyright and privacy protection 18 See https://artificialintelligenceact.eu/ai-act-implementation-next-steps/ . .

The text of the AI Act is available at  http://data.europa.eu/eli/reg/2024/1689/oj .

3 Elements of an EU AI strategy 

In summary, the EU’s current approach to AI is based on catching up on AI hardware and infrastructure, while omitting the complementary business model components and not addressing high regulatory uncertainty and compliance costs. Such an approach is unlikely to solve the fundamental AI competitiveness problem because of the shortcomings set out in the previous section. To address these shortcomings, the EU strategy should include the elements we set out here. Overall, it would be a mistake for the EU to try to play the US at its own game on AI – to reach the AI technology frontier and develop its own AI capacities. Instead, the EU can thrive with smaller models to help firms implement AI-driven services. It does not need to reach the AI technology frontier to accelerate AI-driven productivity growth.

3.1 Facilitate collaboration agreements

Complementary inputs and business ecosystems cannot be created by regulation or public money. They need to grow organically. Competition authorities are taking a close look at collaboration agreements between startups and big tech firms, sometimes rightly so because they may contain exclusivity clauses that distort competition. At the same time, these collaboration agreements and even mergers are necessary to provide the complementary inputs that AI start-ups require. Short of exclusionary contractual clauses, such agreements and mergers should be allowed to go through. Rather than cutting off startups from the complementary inputs they need, EU regulators should focus on solving the missing market failure in private equity markets (as advocated by Draghi, 2024).

3.2 Pro-innovation implementation of the AI Act

The AI Office within the European Commission is in charge of implementing the AI Act, including by designing implementation guidelines and standards (see Box 1). The office should have a razor-sharp focus on pro-innovation implementation and enforcement of the AI Act, minimising compliance costs and navigating the potential pitfalls of strict enforcement. Strict enforcement of existing EU copyright and privacy law is likely to create significant obstacles for AI industries in the EU. The AI Office will have to define an appropriate trade-off between private rights, including the protection of copyright and privacy, and the need to support the development and use of AI services for the benefit of society as a whole. 

3.3 Productivity growth below the AI technology frontier

Apart from trying overcoming these market and regulatory failures through regulatory reform, rather than subsidies, what can the EU do set up a pro-active and pro-competitive AI strategy? Would the EU be better off trying to reach the AI technology frontier, or can it prosper below the frontier? 

Because of delays in AI productivity uptake (Brynjolfsson et al , 2020), most productivity growth will take place below the frontier of the latest generation of GenAI models. Much of the roll-out of AI as a general-purpose technology across the economy will come from derived, smaller and more specialised AI models that can be trained and run at far lower computing costs 19 See Maarten Grootendorst, ‘A visual guide to quantization’, 22 July 2024, https://newsletter.maartengrootendorst.com/p/a-visual-guide-to-quantiza… . . AI applications that can retrieve data in real-time from various sources to respond to user queries will become an important workhorse for industrial applications (Lewis et al , 2024). The CEO of SAP, one of Europe’s leading AI applications companies, has argued in favour of smaller AI models 20 Stephen Morris, ‘SAP chief warns EU against over-regulating artificial intelligence’, Financial Times, 1 October 2024, https://www.ft.com/content/9db8fe6d-3f8a-4886-a439-c23faf459c23 . Even Chinese AI model developers are now turning towards smaller and cheaper AI models. See Eleanor Olcott, ‘Chinese AI groups get creative to drive down cost of models’, Financial Times, 18 October 2024, https://www.ft.com/content/0a6da1bb-2bda-40f3-9645-97877eb0947c . . The focus should be on specialised models designed for specific industrial tasks. This can be done with freely accessible open-source AI models or models that are readily available on the market. It requires access to another type of specialised hardware for inference purposes, not so much for GenAI model training.

3.4 Overcoming complementary market failures

Teece (1980) argued that economies of scope, rather than economies of scale, can be an important source of economic efficiency gains through the re-use of underutilised production factors for other purposes and/or by other parties. However, re-use often fails because markets for complementary inputs are missing or face many barriers, for example because parties cannot agree on a contract to share complementary inputs. This is especially the case in digital services such as AI, where the costs and benefits from combining complements are hard to measure and price (Teece, 2020, 2024). Closer vertical integration is often easier.

Renda’s (2024) observation that European AI patent holders and skilled AI researchers move to the US fits with this story of failing complementary inputs markets. They are looking for complementary inputs that are missing in the EU. EU private equity and venture capital markets for start-ups are underdeveloped. The EU is not home to big tech firms with hyperscale cloud computing infrastructure and business ecosystems that can readily absorb AI services in existing large-scale business models.

Building AI model training infrastructure in the EU is not sufficient to solve that problem; several other complementary inputs will still be missing. The EU could gradually build up markets for complementary inputs by focusing on smaller, derived and specialised AI models that do not require hyperscale infrastructure and business ecosystems and could still earn a decent rate of return for patents and skilled researchers. Smaller venture capital funds and private equity could gradually move into that market. Medium-sized EU cloud service providers could expand their infrastructures and services to accommodate smaller AI models and inference operations.

3.5 Geopolitical dependency

In the current geopolitical security setting, can the EU and US be considered as a single and trustworthy AI market, or are they two separate markets? 

Admittedly, pursuing economic efficiency below the AI frontier would come at the risk of leaving EU AI industrie to some extent dependent on GenAI frontier models developed and/or hosted by US big tech firms. Dependence would only be limited to the extent that many models are freely available in at least partially open-source formats. This raises the question of whether the EU and US can be perceived as a single and trustworthy AI market, or as two potentially separate markets? Here, we take an economic look at that geopolitical question.

Trying to reach the AI frontier is extremely costly (Martens, 2024a) and also requires global market scale. A simple back-of-the-envelope calculation shows that combined GDP of the EU, US and the advanced economies is required to amortize a €1 trillion annual investment in state-of-the-art AI models, a figure that could easily be reached in the next years 21 Assume that digital firms represent 10 percent of GDP and jointly invest €1 trillion in AI models, and that this triggers a 3 percent productivity increase in the remaining 90 percent of the economy, or a 2.7 percent increase in total GDP. To reach break-even on a €1 trillion AI investment would require a GDP of €72 trillion at a 2.7 percent return, assuming that 50 percent of all productivity gains accrue to the digital firms that invest in AI and the rest is dissipated across society. . Collaboration between the EU and US would enable a continuation of that thriving and highly competitive AI industry. Fragmenting the market would put an economic break on that activity. It is not in the interests of either the US or the EU to do this.

In case of fragmentation, the EU would want to build at least some independent AI hyperscale infrastructure to train GenAI models. It will have to bear most of the cost of that infrastructure as a subsidy because, with a reduced market size and in the absence of access to global business ecosystems as AI services outlets, it will be difficult to earn a sufficient rate of return on that fixed-cost investment. The subsidy would be the price for geopolitical independence.

Biglaiser G., J. Crémer and A Mantovani (2024) ‘The Economics of the Cloud’, Working Paper 1520, Toulouse School of Economics

Brynjolfsson, E., D. Rock and C. Syverson (2020) ‘The productivity J-curve: how intangibles complement general purpose technologies’, Working Paper 25148, National Bureau of Economic Research

Cottier, B., R. Rahman, L. Fattorini, N. Maslej and D. Owen (2024) ‘The rising costs of training frontier AI models’, mimeo, available at  https://arxiv.org/abs/2405.21015

Demirer M., D. Jiménez-Hernández, D. Li and S. Peng (2024) ‘Data, Privacy Laws and Firm Production: Evidence from the GDPR’, Working Paper 2024-02, Federal Reserve Board of Chicago

Draghi, M. (2024) The future of European competitiveness: Part A, a competitiveness strategy for Europe , European Commission

Eckert, D. (2024) 40 Years of European Digital Policies: Forgotten Lessons , Springer

Ennis, S. and B. Evans (2024) ‘Cloud Portability and Interoperability under the EU Data Act: Dynamism versus Equivalence’, mimeo

European Commission (2024) ‘Communication on boosting startups and innovation in trustworthy artificial intelligence’, COM/2024/28 final

Goldberg, S.G., G.A. Johnson and S.K. Shriver (2024) ‘Regulating privacy online: An economic evaluation of the GDPR’, American Economic Journal: Economic Policy 16(1): 325-358. 

Jia, J., G. Zhe Jin and L. Wagman (2021) ‘The Short-Run Effects of the General Data Protection Regulation on Technology Venture Investment’, Marketing Science  40(4): 593-812

Hobbhahn, M., L. Heim and G. Aydos (2023) ‘Trends in Machine Learning Hardware’, EpochAI , 9 November, available at  https://epochai.org/blog/trends-in-machine-learning-hardware  

Lewis, P., E. Perez, A. Piktus, F. Petroni, V. Karpukhin, N. Goyal … D. Kiela (2024) ‘Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks’, mimeo, available at https://arxiv.org/abs/2005.11401

Longpre S., R. Mahari, A. Lee, C. Lund, H. Oderinwale, W. Brannon … S. Pentland (2024) ‘Consent in Crisis: The Rapid Decline of the AI Data Commons’, mimeo, available at  https://arxiv.org/abs/2407.14933

Martens, B. (2024a) ‘The tension between exploding AI investment costs and slow productivity growth’, Working Paper 18/2024, Bruegel 

Martens, B. (2024b) ‘Why artificial intelligence is creating fundamental challenges for competition policy’, Policy Brief  16/2024, Bruegel

Peukert, C., S. Bechtold, M. Batikas and T. Kretschmer (2022) ‘Regulatory spillovers and data governance: Evidence from the GDPR’, Marketing Science 41(4): 746-768

Renda, A. (2024) ‘Towards a European large-scale initiative on AI: what are the options?’ CEPS In-depth Analysis 2024-11, Centre for European Policy Studies

Teece D.J. (1980) ‘Economies of scope and the scope of the enterprise’, Journal of Economic Behaviour and Organization  1(3): 223-247

Teece, D.J. (2020) ‘Innovation, governance, and capabilities: implications for competition policy’, Industrial and Corporate Change 29(5): 1075-1099 

Teece D.J. (2024) ‘Strategic Management Perspectives on Competition Policy in the Digital Age’, International Journal of the Japan Association for Management Systems 16(1): 63-67

Vaswani, A., N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. Gomez, L. Kaiser and I. Polosukhin (2017) ‘Attention Is All You Need’, mimeo, available at  https://arxiv.org/abs/1706.03762

Venkataramani, S., X. Sun, N. Wang, C.-Y. Chen, J. Choi, M. Kang … K. Gopalakrishnan (2024) ‘Efficient AI System Design With Cross-Layer Approximate Computing’, Proceedings of the IEEE 108(12): 2232-2250

About the authors

Bertin martens.

Bertin Martens is a Senior fellow at Bruegel. He has been working on digital economy issues, including e-commerce, geo-blocking, digital copyright and media, online platforms and data markets and regulation, as senior economist at the Joint Research Centre (Seville) of the European Commission, for more than a decade until April 2022.  Prior to that, he was deputy chief economist for trade policy at the European Commission, and held various other assignments in the international economic policy domain.  He is currently a non-resident research fellow at the Tilburg Law & Economics Centre (TILEC) at Tilburg University (Netherlands).  

His current research interests focus on economic and regulatory issues in digital data markets and online platforms, the impact of digital technology on institutions in society and, more broadly, the long-term evolution of knowledge accumulation and transmission systems in human societies.  Institutions are tools to organise information flows.  When digital technologies change information costs and distribution channels, institutional and organisational borderlines will shift.  

He holds a PhD in economics from the Free University of Brussels.

Disclosure of external interests  

Declaration of interests 2023

Related content

The tension between exploding AI investment costs and slow productivity growth

This working paper explores the tension between rapidly increasing artificial intelligence investment costs and the slower pace of productivity growth

Draghi disappoints on digital

The Draghi report’s emphasis on hardware and telecoms is out of touch with modern digital developments

Memo to the commissioner responsible for digital affairs

Europe‘s Digital Future. Lessons from the past

How can Europe learn from its past digital strategies to reduce dependence on the US and Asia and better position itself for the digital age?