5. Conclusion

AI evaluation tests aim to estimate the use value, performance, dangers to the safety of goods and people, and positive or negative impact on the environment of these intelligent systems, including their impact on societies and individual lifestyles (socio-economic consequences, ethical, legal, sociological issues, etc.). The organization adopted for these tests, in the form of campaigns, challenges or competitions, depends on the target objective: orientation of R&D efforts, characterization of pre-market performance, benchmarking, voluntary and regulatory certification, etc.

They represent a new challenge linked to the wide potential for deployment of these intelligent technologies in our society. However, they come up against the specific characteristics of AI: intelligent systems do not behave like other information systems. Due to the underlying technologies,...