Researchers are getting tired of OpenAI gatekeeping breakthroughs
Mathematicians and scientists are growing frustrated as OpenAI handpicks expert shepherds to control the release of its latest breakthroughs.
Stock photo for illustration only, not from the actual event
- OpenAI announced major breakthroughs in math, computer science, and physics.
- The company deployed expert shepherds to decide how findings are shared.
- External researchers are expressing frustration over limited access and vetting barriers.
On Monday, OpenAI announced impressive breakthroughs across mathematics, computer science, and physics, accompanied by what the company described as an evolving approach to communication. However, researchers who spoke with The Verge are expressing growing exhaustion regarding the artificial intelligence lab's ongoing conduct.
This friction follows a turbulent period between OpenAI and the mathematics community, who are still dealing with hundreds of preprints dumped unceremoniously at the beginning of October. This time, OpenAI stated it is collaborating with expert shepherds to determine the optimal way to share its latest scientific advancements.
All three research results tackle some of the most consequential unsolved problems in their respective fields, combining prior extensions with entirely new ground. Based on guidance from OpenAI's designated shepherds, each finding utilizes a distinct release strategy:
- Mathematics: OpenAI published the raw AI-generated output, allowing researchers to independently organize efforts to explain the results.
- Computer Science: The company will assemble an external group of researchers specifically to explain the finding.
- Physics: The material has not been released at all, with OpenAI deferring entirely to an external expert group regarding how to handle the documents.
OpenAI's handling of these discoveries highlights the delicate tension between maintaining safety in deploying advanced AI models and the scientific community's demand for radical transparency. By appointing internal-facing shepherds, the company aims to curate complex outputs safely, yet this gatekeeping fuels anxiety that major tech labs wield disproportionate control over the direction of foundational scientific research.
Stock photo for illustration only, not from the actual event
Given these varying disclosure levels and the sheer complexity of evaluating such work, independent researchers face significant hurdles in fully judging the findings. Even with complete information, proper evaluation demands substantial time. Nonetheless, experts speaking to The Verge emphasized that the results are serious, describing them variously as huge, major breakthroughs, and significant.
"It doesn’t make any sense that first only a few distinguished scientists can look at them."
Constantin Kogler
Despite genuine scientific excitement, researchers' enthusiasm remains tempered by exasperation over OpenAI's management style—a familiar sentiment following months of exhausting AI-generated disclosures. Many scientists were eager to dive straight into the rigorous task of scrutinizing the lab's findings, questioning why OpenAI introduced additional barriers by handpicking oversight committees.
Constantin Kogler, a researcher at the Institute for Advanced Study, welcomed the manageable batch of three results compared to the hundreds previously dropped. However, he questioned the restriction of early access to a select group. Meanwhile, Andreas Karch, a physics professor at the University of Texas at Austin, criticized the practice, stating that large labs already hold excessive power and should not dictate who gets access through a handpicked process.
Source: The Verge
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment