Satya Nadella and the three stages of scientific truth
We recorded 6 claims from this post. 2 were checked against evidence. 4 can’t be checked. 1 more is held. Read the post at its source.
Claims from this post (6)
In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.
Assessment: Can’t be checkedSaid
Forecast
“The point was that we probably needed new architectures.”
Our reading: Marcus restates the point of his 2022 essay: AI probably needed new architectures beyond scaling LLMs.
No horizon
Qualifier: probably
Assessment: Can’t be checkedSaid
Observation
“In recent weeks, the third stage has finally arrived.”
Our reading: In recent weeks, as of November 2024, Marcus's 2022 conjecture has reached the stage of being regarded as self-evident, with Microsoft CEO Satya Nadella's remarks as an example. The conjecture is that LLM scaling laws are empirical observations that may not hold and that new approaches would be needed.
No horizon
Scope: in recent weeks
Assessment: SupportedSaid
Observation
“Many other Tech CEOs and others, including Marc Andreeseen (A16Z), Alexandr Wang (ScaleAI), have also used the phrase “hitting a wall” and the idea of diminishing returns in the last couple weeks.”
Our reading: In the last couple of weeks before November 23, 2024, many other tech CEOs and others, including Marc Andreessen and Alexandr Wang, used the phrase "hitting a wall" and the idea of diminishing returns.
No horizon Evidence as of
Limitation: Andreessen did not use the exact phrase 'hitting a wall'; the claim is scored on the group reading.
Assessment: Can’t be checkedSaid
Causal claim
“What’s at stake? Science gets delayed.”
Our reading: Resistance to criticism of scaling slowed scientific progress in AI.
No horizon Evidence as of
Limitation: Causal counterfactual claims of this kind cannot be scored from public records.
Assessment: Can’t be checkedSaid
Causal claim
“Almost a half decade lost, when we could have been seriously investing in other approaches and might have made genuine progress towards robust, trustworthy artificial intelligence.”
Our reading: Focusing on LLMs cost AI research almost half a decade. Investing in alternatives might have produced real progress toward robust, trustworthy AI.
No horizon
Qualifier: if other approaches had been seriously funded
Assessment: UnresolvedSaid
Observation
“Hundreds of billions of dollars were invested on the assumption that I was wrong, largely to the exclusion of investment in other approaches.”
Our reading: Hundreds of billions of dollars went into LLM scaling on the assumption that scaling would keep working, and alternative approaches got little investment.
No horizon Evidence as of
Limitation: The capex figure is a projection made in November 2024 for the full year, not a realized total.
Held from this post (1)
These records did not pass every check, so nothing about them is published. Held is not a finding.