This week, I'm examining the agreement between Le Monde and OpenAI, as well as OpenAI's presentation before the British Parliament. These developments highlight the crucial value of synergy between human and artificial intelligence in ensuring the integrity of training data.The essence of this partnership echoes the "Garbage In, Garbage Out" concept, illuminating that the precision of the output is inherently linked to the quality of the input data. This connection underscores the imperative of enriching AI algorithms with verified and precise data provided by human experts to avoid any risk of AI performance degradation.Although controversial, this agreement is structured to position AI as a complement rather than a substitute to editorial teams, thereby affirming the indispensable role of humans in creative content. This principle reiterates that, although AI can transform or eliminate certain roles, emphasis should be placed on non-AI-based strategies to preserve a wide range of options for the future.OpenAI's stance at the British Parliament, which acknowledges the necessity of using copyright-protected content for training current AI models, invites deep reflection on the balance between such data usage and the respect for intellectual property norms, while highlighting the fundamental importance of human contributions to AI evolution.To conclude, this agreement, along with other similar initiatives with Prisa in Spain and Springer in Germany, heralds an era of specialized AIs trained on high-quality human data, aiming for unprecedented excellence in digital content.  - Dimitri Daniloff

House of Lords Communications
Le Monde