OpenAI's GPT-5.6 Sol Pro disproved a longstanding conjecture about the Benjamini-Hochberg statistical method in under two hours, where previous AI models and human researchers had failed for decades.
A University of Pennsylvania statistics professor leveraged GPT-5.6 Sol Pro to disprove a central open conjecture about the Benjamini-Hochberg method—a fundamental statistical technique used across scientific research—in approximately 90 minutes.
The achievement marks a significant milestone in AI problem-solving. The model's predecessor, GPT-5.6, could not identify a solution even after 20 hours of computation, underscoring the incremental improvements in the latest generation.
The Benjamini-Hochberg method, developed in 1995, is widely used to control false discovery rates in multiple hypothesis testing. Despite three decades of investigation by statisticians, a key conjecture about the method's properties remained unproven—until now.
The Solution's Nature
The disproof reportedly combines known mathematical and statistical methods in a novel configuration. This detail introduces a crucial distinction: while GPT-5.6 Sol Pro produced a breakthrough result, the underlying techniques were already established within mathematical literature.
This raises the enduring question at the intersection of AI and mathematics: whether large language models can generate genuinely original knowledge or primarily synthesize and recombine existing information in new ways.
Implications
The result demonstrates that AI systems can pattern-match across vast repositories of mathematical knowledge and apply those patterns to unsolved problems with efficiency exceeding human researchers' capabilities. Whether this constitutes "genuine" discovery depends on how one defines knowledge creation.
For the statistics community, the disproof itself matters less than the methodology. If AI can reliably tackle open conjectures by finding novel combinations of established techniques, it becomes a valuable tool for accelerating research across mathematics and related fields.
The development also illustrates AI's growing role in computational problem-solving, though questions persist about whether such systems can independently generate fundamentally new mathematical insights or concepts.
A developer who scraped artwork for AI training is now collaborating with Cara, a creator platform designed to prevent unauthorized AI data collection, as the service faces ongoing attacks from trolls attempting to breach and publish its data.
Installing a large language model on your personal computer creates a private digital assistant without uploading data to external servers. This approach keeps your information secure while giving you full control over the AI.
LAION has published Big Video Dataset (BVD), an open collection containing 80 million videos and 10 million hours of footage designed for AI research. Models trained on BVD outperform the previous benchmark InternVid by up to 2.1 percentage points.
As generative audio tools improve, AI-created music floods the internet. Some creators deny using the technology until public pressure forces them to confess.