MANY WAYS
Augmenting human minds: artificial intelligence and big data in risk assessment
Risk assessment has reached its limits in the ability to be executed in a timely manner: recruiting relevant experts is increasingly demanding, while their physical capacity to identify, search, read, appraise, and integrate the exponentially growing amount of data in a structured way is stretched to a breaking point. Automation of the process and the application of artificial intelligence (AI) are anticipated to play an increasingly key role in future risk assessments. AI’s potential contribution to the further advancement of risk assessment is two-fold: (1) identify and mine the available data and expertise that are relevant for a given risk assessment; and (2) execute parts of the risk assessment. This session will explore the role that AI can play in the conduct of future risk assessments and analysis of big data, what challenges its implementation may entail in a regulatory context, and how to transition to trustworthy AI systems for regulatory science.
Vision
AI and automation hold enormous potential to: (1) speed up the conduct of risk assessments; (2) keep up with the exponentially growing amount of data potentially relevant for risk assessments; and (3) manage and exploit big datasets. This would enable an “incremental” approach to future risk assessments; one that takes newly produced evidence into account and remains up to date in real-time. Consequently, to deliver more reactive risk assessments and ensure preparedness for future risk analysis needs, EFSA aims to implement AI approaches by 2027.
EFSA also envisages to adopt human centric AI techniques, as it is expected that humans and AI will operate together in complementarity for future risk assessments.
Background – Challenges and opportunities
In December 2018, the European Commission, together with EU Member States, developed a Coordinated Plan tailored to: (1) maximise the impact of both national and European investments in AI; and (2) foster synergies and cooperation on AI across the EU. The European Commission also appointed a group of experts (i.e., High-Level Expert Group on AI) to provide advice on its AI strategy.
In 2020, the European Parliament created a special committee on Artificial Intelligence in a Digital Age (AIDA) that seeks to develop a long-term strategy on AI and how AI can be used in the EU. Among other aspects, it investigates the impact of AI on the EU economy.
Overall, the use of AI technology is strongly recommended, as it may address global challenges such as climate change, transport, safety, and cybersecurity. Moreover, the volume of data and scientific evidence is growing exponentially, requiring complementary approaches for evidence identification, search, reading, appraisal, and integration.
The approach to AI currently explored and proposed at EU level is characterised by a highly human-centric AI component. This is because new tools and human-centric approaches, including AI, are required to take advantage of large datasets, including those considered for risk assessment purposes.
However, building trustworthy AI systems in support of future risk assessments will entail various challenges (such as robustness, generalisation, explainability, transparency, reproducibility, fairness, privacy preservation, alignment with human values, and accountability) that need to be overcome. Besides challenges, the use of AI and Big Data offers unique opportunities. It could speed up risk assessments, improve the quality of risk assessments (e.g., limit human error), enable discovering new patterns in the data landscape that are undetectable by humans, and keep risk assessments up to date despite the exponential increase of evidence.
Scope and objectives
The rapidly growing amount of available evidence to digest, the recruitment of relevant experts for the conduct of risk assessment, and their physical capacity to identify, search, read, appraise, and integrate large datasets in a structured way are increasingly challenging, requiring new tools and approaches. The use of AI holds great promise to overcome these challenges. Therefore, the thematic session will explore the role that AI can play in the conduct of future risk assessments and analysis of big data, what challenges its implementation may entail in a regulatory context, and how to transition to trustworthy AI systems for regulatory science.
The main objectives of the thematic session are to:
- Explore whether human expertise needed in risk assessment can be replaced partially or fully by AI-powered models.
- Discuss which parts of risk assessment processes are open to automation.
- Identify which risk assessment-related processes will continue to require human expertise.
- Explore whether future risk assessments will follow a “human-augmented” approach in which humans and AI operate together in complementarity.
- Make recommendations on how to develop trustworthy AI systems for regulatory science.
| Time | Duration | Talk |
|---|---|---|
| 09:00 | 10 |
Opening and welcome Ilaria Del Seppia, European Medicines Agency (EMA) | Guy Van Den Eede, European Commission |
Part I - Challenges and opportunities associated with the use of AI in future risk assessments
| Time | Duration | Talk |
|---|---|---|
| 09:10 | 20 |
Opportunities of Big Data and AI in Food Safety Risk Assessment Hans Marvin, Wageningen Food Safety Research (WFSR) |
| 09:30 | 20 |
Watching the watchmen: testing AI systems intended for risk assessment Raluca Crisan, Etiq AI |
| 09:50 | 15 |
Q&A Ilaria Del Seppia, European Medicines Agency (EMA) | Guy Van Den Eede, European Commission |
| 10:05 | 30 |
Coffee break |
Part II - Considerations and recommendations for a concrete adoption of AI in future risk assessments
| Time | Duration | Talk |
|---|---|---|
| 10:35 | 75 |
Panel discussion moderated by Jennifer Baker Ivana Bartoletti, Wipro | Rens Van de Schoot, Utrecht University | Akos Jozwiak, University of Veterinary Medicine Budapest | Markus Lipp, Food and Agriculture Organization of the United Nations (FAO) |
| 11:50 | 10 |
Wrap up and concluding remarks Ilaria Del Seppia, European Medicines Agency (EMA) | Guy Van Den Eede, European Commission |