Centaur, the AI Claiming to Simulate the Human Mind: Cognitive Revolution or Technological Illusion?

The Centaur project, developed by a consortium of researchers and companies specializing in artificial intelligence, is sparking a passionate debate within the scientific community. According to a survey by Courrier International, this AI aims to model the functioning of the human mind in an unprecedented way, by combining symbolic learning and deep neural networks. Centaur seeks not only to understand but also to replicate the logic, intuition, and creativity characteristic of human cognition. A goal that is as fascinating as it is concerning, given the ethical, social, and scientific implications of such a feat could disrupt our relationship with technology.

An Unprecedented Technological Gamble: Towards a Hybrid Intelligence

The Alliance of Symbolic and Neural

Centaur stands out from other current AI models with its hybrid approach. While most large language models (LLM) rely solely on statistical learning, Centaur integrates a symbolic engine that allows it to reason like an expert system, combined with a neural model capable of assimilating information and generalizing from massive data. According to the designers, this combination would enable the AI to go beyond mere linguistic imitation to acquire a form of semantic understanding.

In practice, Centaur would be capable of solving complex problems, manipulating abstract concepts, and connecting ideas from different contexts, skills that approach human reasoning. The initial results, shared by the project team, show an ability to find creative solutions to logical puzzles, surpassing current models like GPT-4 or Claude on certain specific benchmarks.

A Step Towards Understanding Mental Processes

Researchers claim that Centaur does not just produce relevant answers but also offers insight into how it reaches its conclusions. This "cognitive traceability" could allow scientists to better understand certain mental processes, such as decision-making or the formation of symbolic representations, opening new perspectives for neuroscience and cognitive psychology.

Between Scientific Promise and Potential Pitfalls

The Risk of Anthropomorphic Illusion

Despite these advances, several experts emphasize that one should not confuse simulation with real understanding. Centaur, like any AI, does not "think": it performs complex calculations but does not feel emotions or intentions. According to an article cited by Courrier International, the danger is attributing to Centaur human faculties it does not possess, a bias called anthropomorphism, which could mislead users and lead to excessive trust in its responses.

This risk is even greater if Centaur is used in sensitive areas such as medical advice, justice, or politics, where a misinterpretation could have serious consequences. The researchers themselves acknowledge the need to train users to understand the limits of AI, to avoid considering it as a substitute for human intelligence.

Legal and Intellectual Property Issues

The Centaur project relies on massive training of its model from textual, visual, and auditory data, some of which may be protected by copyright. This use raises questions about the legality of exploiting these contents, as well as the ownership of the productions generated by the AI: who owns the original ideas or solutions it proposes? These questions are all the more pressing as Centaur aims to become a research and innovation tool for companies, which will need to ensure they do not infringe on others' intellectual property.

With Centaur, AI ventures into uncharted territory: modeling the mechanisms of human thought. If the project delivers on its promises, it could revolutionize our understanding of cognition and open new applications in medicine, education, or fundamental research. But without safeguards, this advance also risks creating dangerous illusions about the real capabilities of AI.

To regulate the use of Centaur, companies and institutions will need to:

• contract agreements specifying rights on the content used for training;
• include clauses on liability in case of AI errors or damages caused by its use;
• implement a transparency policy on the model's functioning and limitations.

These measures will allow the potential of Centaur to be harnessed while protecting users and rights holders.

This initiative raises an essential question: should there be international regulation of AIs claiming to simulate human thought? The coming years will be decisive in establishing a clear framework, to make the most of these technologies without sacrificing user trust and safety.