“The transformer-based architectures like the kind used in OpenAI’s GPT models are gigantic math-machines that know nothing, and will never, ever create the Artificial General Intelligence (AGI) that Sam Altman has been claiming it will.”
Our reading:
Transformer-based models like OpenAI's GPT know nothing and will never create AGI.
Assessment: Unresolved
What happened
Did the forecast come true?
Outcome: Unresolved
We checked, and the evidence so far doesn’t settle it.
Judged on a plain reading. They named no measure, so we used the reading a typical informed reader would take. It is spelled out under “How we judged it”.
Evidence up to . Checked .
Why
On September 3, 2026, ARC Prize reported that Astra used fewer actions than the median tested human on 96% of ARC-AGI-3 levels. [e2] Its evaluation does not establish AGI. [e2] The AGI Society's September 2026 assessment says there is no generally accepted empirical test for AGI and does not accept the published Astra evidence as establishing it. [e1] Neither source provides a qualifying transformer-based AGI demonstration or an accepted impossibility result, leaving the forecast unresolved. [e1, e2]
Caveats
- Benchmark results alone do not establish broad human-level general intelligence.
- The packet does not establish that future transformer-based systems cannot produce AGI.
- The claim leaves AGI, the boundary of transformer-based architectures, and 'know nothing' undefined.
- This assessment addresses the AGI forecast and does not separately resolve 'know nothing'.
Evidence (2 items)
Context only: Artificial General Intelligence Society,
There is no generally accepted empirical test for AGI.
Why it matters: The Society's assessment does not accept the published Astra evidence as establishing AGI. This explains why declarations of achievement do not establish the required counterexample; it does not prove future impossibility.
- Publisher
- Artificial General Intelligence Society
- Kind of source
- Secondary summary
- About events on
- Published
Context only: ARC Prize,
96% of levels.
Why it matters: This is material evidence of capability on unfamiliar interactive tasks. The evaluator nevertheless says that benchmark saturation is not proof of AGI and does not claim Astra is AGI.
- Publisher
- ARC Prize
- Kind of source
- Independent measurement
- About events on
- Published
How we judged it
- Kind of claim
- Forecast
- Deadline
- None given. We don’t add one.
Where we had to interpret
- The testable forecast is that transformer-based architectures cannot ever produce AGI. No deadline is supplied.
- The ordinary reading of AGI is broad human-level general intelligence, demonstrated across unfamiliar tasks rather than established by one benchmark.
- Broad independent acceptance and credible disclosure of the system's architecture would be required for a decisive counterexample.
- The phrase 'know nothing' supplies no separate test of knowledge or understanding. It is retained as an ambiguity, not converted into a claim that models cannot answer factual questions.
Right if: An accepted impossibility result establishing that the stated architecture class cannot meet broadly accepted AGI criteria. Continued failure to demonstrate AGI is insufficient.
Wrong if: A publicly demonstrated, independently evaluated transformer-based system meets broadly accepted AGI criteria.
We wrote this test after the outcome was known.
What would settle it
A publicly demonstrated, independently evaluated transformer-based system broadly accepted as AGI, with credible architecture documentation; or an accepted impossibility result covering the architecture class and relevant AGI criteria.
We found 28 claims in this post. See them all.
Full record
Source and dates
Where's Your Ed At, The Shareholder Supremacy, paragraph 121.
- Said
- We read the source
- live page, full text, public
- Evidence up to
- Checked
We read this wording on 2026-10-28. The post may have been edited after 2024-07-01.
Where it was said (1)
Said once in the posts we read.
: Where's Your Ed At, The Shareholder Supremacy
The transformer-based architectures like the kind used in OpenAI’s GPT models are gigantic math-machines that know nothing, and will never, ever create the Artificial General Intelligence (AGI) that Sam Altman has been claiming it will.
- Where
- paragraph 121
- Published
- 2024-07-01
- We read it
- , live page
- Access
- full text, public
History
Versions
- Current: Unresolved. Evidence up to 2026-10-28.
What happened to this record
PublishedBy us
Published after both AI reviews and the final check passed.
Held backBy us
The final check failed: a cited source didn’t say what the write-up said.