OpenAI announces three 'major' science results and names outside experts to vet them
On 26 October OpenAI announced AI-produced results in mathematics, computer science and physics, including a claimed refutation of the Exponential Time Hypothesis. It said outside experts, called "shepherds", would help verify and explain them. None of the results has been independently confirmed.
1 / 2
The story, neutrally told
Mixed · 2On Monday 26 October OpenAI announced three "major" discoveries in mathematics, computer science and physics. Hindustan TimesN “OpenAI announced on Monday, October 26, that it has made three "major" discoveries in physics, mathematics and computer science.” Read at Hindustan Times ↗ The VergeLC “OpenAI is back with three more “major” discoveries in math, computer science, and physics.” Read at The Verge ↗ Mixed · 2In computer science, OpenAI says it has a deterministic algorithm that counts satisfying assignments of fixed-width CNF formulas in subexponential time and polynomial space. It says the 3-SAT case refutes the Exponential Time Hypothesis (ETH), which was widely believed to be true. The VergeLC “The case of 3-SAT refutes the Exponential Time Hypothesis.” Read at The Verge ↗ Hindustan TimesN “long believed to be true, is actually false” Read at Hindustan Times ↗ Left · 1In mathematics, OpenAI says its model proved the local smoothing conjecture in every dimension. It says this also resolves the Kakeya conjecture and Stein's restriction conjecture in all dimensions. The VergeLC “produced a proof of the local smoothing conjecture in every dimension and therefore also resolve the Kakeya conjecture and Stein’s restriction conjecture in all dimensions.” Read at The Verge ↗
Left · 1The Verge notes that NYU professor Hong Wang was awarded a Fields Medal a few months ago for resolving the Kakeya conjecture in three dimensions. The VergeLC “NYU professor Hong Wang was awarded a prestigious Fields Medal by the International Mathematical Union for resolving the Kakeya conjecture in three dimensions.” Read at The Verge ↗ Mixed · 2The physics result has not been released in full. OpenAI describes it as a "mathematically rigorous existence proof of certain interacting quantum field theories containing gauge fields and fermions in four dimensions, including Quantum Chromodynamics (QCD) and a unified completion of the Standard Model of particle physics." The VergeLC “The physics information has not been released in full” Read at The Verge ↗ Hindustan TimesN “The discovery in Physics has not been fully disclosed” Read at Hindustan Times ↗ Centre · 1What OpenAI says is new is the use of external experts as "shepherds", who organise groups to understand the results. OpenAI said it was "experimenting with a new approach" to turn raw research output into scientific understanding. Hindustan TimesN “We reached out to external experts and asked them to act as shepherds, organizing groups to understand the new results.” Read at Hindustan Times ↗
Left · 1David Gross will lead a group including Michael Douglas, Arthur Jaffe, Igor Klebanov and Nikita Nekrasov to shepherd the physics work. The VergeLC “David Gross will lead a group including Michael Douglas, Arthur Jaffe, Igor Klebanov, and Nikita Nekrasov that will shepherd this work.” Read at The Verge ↗ Left · 1For the ETH result, a working group led by Russell Impagliazzo, Ramamohan Paturi and Ryan Williams will examine and explain the proof. Impagliazzo and Paturi, of UC San Diego, were part of the group that proposed the ETH in 1999. At the group's request, the write-up and Lean proof were posted publicly on GitHub immediately. The VergeLC “An independent working group of computer scientists led by Russell Impagliazzo, Ramamohan Paturi, and Ryan Williams will examine and explain the proof.”“who were part of the group that originally proposed it in 1999” Read at The Verge ↗ Left · 1For the mathematics result, OpenAI released the model output directly, as it did on 6 October. The Verge reports that the experts advised this so researchers could work out their own explanations. The VergeLC “We are releasing the model output directly, as we did in our October 6 release.”“they advised it to release the AI output for researchers to work through their own explanations” Read at The Verge ↗
Centre · 1The announcement follows OpenAI's release last month of hundreds of manuscripts claiming solutions to long-standing maths problems. Hindustan Times says the company was criticised for bypassing academic peer review then. Hindustan TimesN “The company faced criticism for bypassing academic peer review when it released hundreds of manuscripts with solutions to long-standing math problems last month.” Read at Hindustan Times ↗ Left · 1The Verge adds that the new approach may have its own problems, which the company admits. It says an AGMAI panel of math advisers assembled by OpenAI said a day earlier, "we recommend that AI companies do not adopt" the practice of designating certain people to receive output first. The panel cited unequal access, borrowed reputation and biased selection. It instead recommended releasing output on scholarly repositories with Lean-formalised proofs, information on how problems were chosen, and funding for people working to understand the results. The VergeLC “the AGMAI panel of math advisers OpenAI put together just said yesterday that “we recommend that AI companies do not adopt” the practices of designating certain people to receive the output first”“citing unequal access, borrowed reputation, and biased selection as a few of the issues” Read at The Verge ↗
Every sentence links to the reporting it rests on. The pill in front of each says where its sources sit: Left, Centre or Right when one side supplies at least half of them, Mixed when they are evenly split. The number is how many outlets it cites.
Left1 outlet
- Framing
- The Verge treats the announcement as a new way of working with academics and sets it against an advisory panel's warning against the same practice.
- Emphasis
- The tension between OpenAI's shepherd model and the AGMAI panel's recommendations; the Fields Medal context for Kakeya.
- Leaves out or plays down
- Does not quote any independent expert's verdict on whether the proofs are correct.
- Charged language
- “major”
Centre1 outlet
- Framing
- Hindustan Times reports the announcement largely through OpenAI's own words, as three claimed breakthroughs and a changed approach after earlier criticism.
- Emphasis
- The three headline claims, the earlier peer-review criticism and the named expert leads.
- Leaves out or plays down
- Does not mention the AGMAI panel's recommendation against designated early recipients. It also does not say that the ETH write-up and Lean proof were posted on GitHub.
- Charged language
- “major”“breakthroughs”
- For example
-
“OpenAI claims three major breakthroughs in physics, mathematics and computer science, including resolving two mathematical conjectures and challenging ETH.” — Hindustan Times
“This time, the company followed a different approach by incorporating academics as "shepherds" to guide the findings.” — Hindustan Times
Right0 outlets
No right outlet in our sources has covered this story yet.
What every side reports
- OpenAI announced on 26 October three results in mathematics, computer science and physics.
- The computer science claim is that the Exponential Time Hypothesis is false.
- The mathematics claim covers the Kakeya and Stein restriction conjectures in all dimensions.
- The physics proof has not been fully released.
- External experts called "shepherds" are named for the computer science and physics work, including Russell Impagliazzo and Ramamohan Paturi, and David Gross.
Where accounts differ
-
Which result had its raw AI output released on the experts' advice
- Left
- The Verge says the experts on the maths problem advised releasing the AI output so researchers could work through their own explanations. It says the computer science group chose to post its write-up and Lean proof on GitHub.
- Centre
- Hindustan Times says the group recommended releasing the raw AI output for the computer science discovery.
OpenAI organisation
OpenAI presents the results as major discoveries. It says it is experimenting with outside "shepherds" to turn raw AI output into scientific understanding, and that the experts advised how each result should be released.
“To convert this raw research data into scientific understanding, in this release we are experimenting with a new approach,” — Hindustan Times
“At the group’s request, we are making the materials public immediately so others can examine them as well.” — The Verge
Left1 article
-
OpenAI announces three big science discoveries and a new way to work with academics
Mixed Reports OpenAI's claims and its new shepherd approach, and balances them with the advisory panel's warning against designated early recipients.
Centre1 article
-
OpenAI announces three ‘major’ science discoveries in math, computer science and physics
Neutral Relays OpenAI's announcement and claims, noting earlier peer-review criticism and the change of approach.
Right0 articles
No coverage yet.
- 26 Oct 23:34 First The VergeLC OpenAI announces three big science discoveries and a new way to work with academics
- 27 Oct 00:07 +33m Hindustan TimesN OpenAI announces three ‘major’ science discoveries in math, computer science and physics
Times are when each article was published, or when we first saw it if the outlet gave no time.
Ask about this story
Loading…