Key learnings from the keynote on generative AI by Hanan Ouazan, Partner and Generative AI Lead at Artefact, at the Adopt AI Summit by Artefact - June 5, 2024

About Hanan Ouazan: Hanan Ouazan has graduated from Ecole Normale Supérieur and Ecole Centrale de Lille in ML and AI. He joined Artefact almost 10 years ago and has been for the past few years the Lead for Generative AI.

Rapid evolution of technologies

Hanan Ouazan discusses the remarkable changes over the past ten years, notably the shift from solution design to production deployment and from feature engineering to prompt engineering, as well as the evolution of interaction with machines through interfaces like ChatGPT.

Growing interest from companies

He highlights that many companies are interested in generative AI, with new products launched daily and rapid innovations from major players, although few have successfully transformed their business with this technology.

“Integrating generative AI means accepting its defaults and setting up the prerequisites for success. We need to reshape our way of working to integrate and not superimpose AI.”
Hanan Ouazan, Partner & Generative AI Lead at Artefact

Challenges to overcome

The main obstacles to adopting generative AI are data quality, the high costs of models (which are decreasing), and legal and regulatory issues, with the AI Act facilitating regulation.

Trust issues

The key to overcoming scaling problems lies in trust in generative AI. Model errors are compared to human “hallucinations,” and the need to control and make machine errors acceptable is emphasized.

Necessary mindset change

To succeed with AI, it is crucial to change our behavior and integrate AI into processes rather than superimposing it. Hanan Ouazan concludes by calling for better understanding and management of AI for successful adoption.

Role of France and research laboratories

Hanan Ouazan highlights France’s strong position in the AI field, with influential research laboratories and a very active open-source community contributing to innovation and the development of new technologies.

Importance of regulation and collaborative initiatives

He stresses the role of regulations like the AI Act in creating a safe framework for innovation and mentions collaborations with companies like Mistral AI to develop solutions aimed at preventing errors and effectively controlling algorithms.