Artificial Intelligence for Innovation Assessment (AI4IA)

About

Innovation is driving most of the transformational changes we observe in the Grand Challenges we face as a society, from environmental protection and adaptation, to healthcare provision passing by work redefinition with the advent of AI. Properly assessing innovation and measuring it is therefore a key skill for any researcher studying these questions. In this course we will build on a rather old tradition and discipline but as facing new assessment challenges ahead.

Course contents and objectives

Innovation agencies, regulators, and research organizations increasingly need to assess concrete innovations in deployment: to understand who is affected, how strongly, and how quickly – and to inform timely regulation and policy responses. Think of innovation such as new platforms for ride or home sharing, social networks, online shopping, massive online courses, AI applications in many domains. All these innovations combine new technologies, new business models and new user’s experiences. All are likely to significantly affect social interactions and social patterns. But – before their “socialization”, we don’t know what is going to happen – precisely because their socialization is the mechanism through which effects and properties will be elicited.

The goal of this course is to help PhD students understand how to assess innovations under uncertainty and in real time, how to measure their economic and social effects empirically, and how to design experiments that accelerate learning while protecting consumers. Further, we will analyze how to leverage artificial intelligence tools to enhance the assessment process and how to critically evaluate the validity and generalizability of assessment findings. This course equips participants with a methodological toolbox in “Innovation Assessment”, and hands-on experience to:

Measure innovation and knowledge flows: using bibliometric and patent data, text-based analysis and modern indicators of novelty, interdisciplinarity and proximity, relying on AI and ML tools for large textual corpora.
Assess the societal and economic effects of specific innovations using accessible causal inference tools on real-world data
Use LLMs as research infrastructure to support (not replace) empirical work (design, prototyping, scenario building) while critically evaluating validity and limitations
By the end of the week, throughout accompanied practical exercises, participants will be able to design a pragmatic innovation assessment protocol for a real innovation case, have a set of re-usable scripts and prompt templates, be able to critically judge when and how to use AI tools in policy-oriented empirical work.

The course will build on two main pillars. On the one hand, an understanding of innovation as a process of social and economic experimentation, where new technologies and business models are deployed in society and their properties are progressively discovered through adoption and use. In such a perspective, activities such as measuring innovation outputs through bibliometric and patent data, tracking spillovers and diffusion patterns, and designing controlled experiments become central to understanding innovation impacts. Innovation assessment takes place largely through empirical analysis, which follows specific methodological conventions for data collection and causal inference, but also requires careful attention to temporal dynamics, external validity, and the interpretation of effects that emerge over time. On the other hand, the course will build on recent advances in artificial intelligence, which have transformed research practice by enabling rapid literature synthesis, hypothesis generation, and even synthetic data creation for policy simulation. These advances lead to new possibilities for innovation assessment, in which researchers can accelerate the detection of innovation effects through AI-augmented analysis while maintaining critical awareness of validity concerns. In such a perspective, researchers strategically combine traditional empirical methods with AI tools to pursue multiple objectives, such as identifying potential negative effects earlier, mapping stakeholder ecosystems more comprehensively, and designing more effective regulatory responses.

Understanding innovation assessment in these terms will help students to better grasp how to apply both classical measurement techniques and emerging AI capabilities, while recognizing the limitations and appropriate contexts for each approach.

The course will be organized in face-to-face lectures and in practical exercises, in which students will work with real innovation datasets, replicate canonical empirical studies, and apply large language models to structured research tasks. It will focus in this respect on three major domains of innovation assessment: measuring innovation through bibliometric and patent analysis, assessing innovation effects in real-world settings, and leveraging AI to accelerate the assessment process.

Course design

Face-to-face lectures in the mornings, practical work in the afternoon:

Hands-on exercises with open data sources (OpenAlex, patent databases);
Replication of empirical studies on innovation measurement and assessment;
Structured application of large language models to research tasks;
Team project developing innovation assessment protocols.

Detailed lecture plan (daily schedule)

Day 1
The Traditional Innovation Framework: measuring innovation through publications and patents. Bibliometric and patent analysis foundations.

Afternoon: Mapping knowledge flows from academic publications to inventions using open data sources.

Day 2
Innovation quality and spillovers: measuring knowledge diffusion and cumulative innovation. The science-invention interface.

Afternoon: Replication exercise examining the relationship between research quality and invention value.

Day 3
Innovation assessment as a new discipline: from ex ante technology assessment to ex post innovation assessment. The role of experimentation and learning under uncertainty.

Afternoon: Empirical exercises on assessment timing (Airbnb case) and external validity (bike-sharing comparison across contexts).

Day 4
AI-augmented innovation assessment: the transformation of research practice through large language models. Implications for accelerating innovation assessment.

Afternoon: Structured application of LLMs to innovation cases; critical validation of AI-generated outputs.

Day 5
Project work: morning preparation of team innovation assessment protocols; afternoon presentations and discussion.

Class materials

Datasets for practical exercises (OpenAlex, patent data, Airbnb and housing price data).
Python notebook templates in Google Colab (browser-based, no installation required).
Prompt templates for large language model exercises.
Reading package of key papers

Course leader

Prof. Dr. Dominique Foray is professor emeritus at the École Polytechnique Fédérale de Lausanne and a member of the Swiss Science Council. He held the Chair of the Economics of Innovation at EPFL from 2004 to 2022. His research focuses on the economics of knowledge and on the economics and policies of innovation. Prof. Dr. Charles Ayoubi is an Assistant Professor at ESSEC Business School. His research focuses on how organizations generate, evaluate, and diffuse innovative ideas, with particular attention to the role of generative AI in innovation and collaboration. Prof. Dr. Omar Ballester is Policy Officer at Swiss National Science Foundation (SNSF) and Adjunct Professor at EHL University of Applied Science Western Switzerland (HES-SO). His work bridges innovation economics and machine learning (AI), with a focus on applied policy questions and evidence informed governance of science and innovation systems

Target group

primarily graduate researchers, PhD researchers, early career researchers

**The Summer School cannot grant credits. We only deliver a Certificate of Participation, i.e. we certify your attendance.**

If you consider using Summer School workshops to obtain credits (ECTS), you will have to investigate at your home institution (contact the person/institute responsible for your degree) to find out whether they recognise the Summer School, how many credits can be earned from a workshop/course with roughly 35 hours of teaching, no graded work, and no exams.

Make sure to investigate this matter before registering if this is important to you.

Prerequisites

Basic familiarity with empirical research methods and data analysis. Ability to read and interpret quantitative studies. No prior programming or AI experience required – all necessary tools and techniques will be introduced during the course. PhD students and early career researchers in social sciences, management, policy studies, and related fields will benefit most from the course

Fee info