OpenAI's GPT-5.6 Solves a Statistical Conjecture in 90 Minutes
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
A Spectacular Breakthrough for GPT-5.6 Sol Pro
A statistics professor from the University of Pennsylvania recently tested OpenAI's GPT-5.6 Sol Pro model, and the results are impressive. In just 90 minutes, this artificial intelligence model managed to solve a statistical conjecture that had remained unresolved for 30 years, related to the Benjamini-Hochberg method.
A Failure for GPT-5.5
The predecessor of this model, GPT-5.5, had attempted to tackle the same problem but was unsuccessful. After 20 hours of analysis, it was unable to find a solution, highlighting the significant advancement represented by GPT-5.6 Sol Pro.
A Recombination of Existing Methods
The solution provided by GPT-5.6 Sol Pro relies on an innovative combination of known methods. However, this raises a crucial question: is AI capable of generating truly new knowledge, or does it merely reorganize the information it has already assimilated? This inquiry is at the heart of the debates surrounding the potential and limitations of artificial intelligence in the field of scientific research.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.