Will AI soon be able to create the AI of tomorrow?

In an article published in Le Temps, Daniel Steck explores the debates surrounding the next stages in the development of artificial intelligence.

Following a warning issued by Anthropic, the developer of the Claude chatbot, the prospect of artificial intelligence becoming capable of developing increasingly advanced models on its own has once again come under the spotlight. Such a scenario raises important questions about humanity’s ability to maintain control over this technology.

According to Daniel Steck, such a development would require the emergence of Artificial General Intelligence (AGI), a form of AI capable of learning, understanding and adapting to any intellectual task in much the same way as a human being. Achieving AGI remains one of the industry's most ambitious goals and helps explain the significant investments currently being made in AI.

Daniel Steck also notes that a voluntary slowdown in this technological race appears unlikely. Beyond competition among major US technology companies, artificial intelligence has become a strategic battleground in the broader technological rivalry between the United States and China.

As AI models continue to advance at a rapid pace, questions surrounding regulation and safeguards are becoming increasingly pressing. Daniel Steck points out that even the United States has recently strengthened certain protective measures aimed at securing critical infrastructure against the new capabilities enabled by AI.

At the same time, beyond some of the more alarmist narratives, he emphasises that leading industry players have been highlighting the ethical and societal challenges associated with artificial intelligence for several years. The potential impact of this technology could profoundly reshape both our economy and society.

Read Daniel Steck’s full analysis in Le Temps (french only)