Andrew Ng on Mandate big AI companies spend 30% of c… | YouCongress

Andrew Ng

Baidu; Stanford CS faculty; founded Coursera and Google Brain

However, it is important to differentiate regulating applications (which we need) vs. regulating the technology (which is ill-advised).

AI Verified source (Dec 11, 2023)

Like

3mo ago

Policy proposals and claims

Mandate big AI companies spend 30% of compute on AI safety

votes Against

Statement relation verification history AI Verified Report this

AI Verified The quote is not directly about a 30% compute requirement, but it clearly argues that regulating AI technology itself is ill-advised. A mandate forcing big AI companies to devote 30% of compute to safety is a technology-level regulation, so the quote implies opposition to the full policy. · YouCongress gpt-5.4-2026-03-05 · 16d ago

Vote inference verification history AI Verified Report this

AI Verified The quote says "regulating the technology" is "ill-advised," and a mandate on how big AI companies allocate compute is a form of regulating the technology. · YouCongress gpt-5.4-2026-03-05 · 16d ago

Establish a UN-led body to oversee compute-intensive AI

votes Against

Statement relation verification history AI Verified Report this

AI Verified The quote clearly opposes regulating AI technology itself: it says regulating applications is needed, but regulating the technology is 'ill-advised.' A UN-led body to oversee compute-intensive AI would be a form of technology oversight, so the quote implies opposition to the full statement. · YouCongress gpt-5.4-2026-03-05 · 16d ago

Vote inference verification history AI Verified Report this

AI Verified The quote says "regulating the technology" is "ill-advised," which clearly cuts against a UN-led body to oversee compute-intensive AI as a form of technology regulation. · YouCongress gpt-5.4-2026-03-05 · 16d ago

Require AI labs to publish safety evaluations before deploying frontier models

votes Against

Statement relation verification history AI Verified Report this

AI Verified The quote is not directly about safety evaluations, but it clearly implies opposition to the statement: requiring labs to publish evaluations before deploying frontier models is a form of regulating the technology itself, which the author says is "ill-advised." · YouCongress gpt-5.4-2026-03-05 · 16d ago

Vote inference verification history AI Verified Report this

AI Verified The quote says "regulating the technology" is "ill-advised," and this proposal is a form of regulating AI technology rather than applications. · YouCongress gpt-5.4-2026-03-05 · 16d ago

Require open source AI models to include safety benchmarks in release notes

votes Against

Statement relation verification history AI Verified Report this

AI Verified The quote clearly contrasts supporting regulation of AI applications with opposing regulation of the technology itself. Requiring open-source models to include safety benchmarks in release notes is a technology-level regulation on model releases, so the quote implies opposition to the full statement. · YouCongress gpt-5.4-2026-03-05 · 16d ago

Vote inference verification history AI Verified Report this

AI Verified The quote says "regulating the technology ... is ill-advised," and this requirement is a technology-level rule for open source AI models. · YouCongress gpt-5.4-2026-03-05 · 16d ago

Mandate reporting of AI training runs above 10²⁶ FLOPs to a designated national or international authority

votes Against

Statement relation verification history AI Verified Report this

AI Verified The quote clearly contrasts supporting regulation of AI applications with opposing regulation of the technology itself as "ill-advised." Requiring developers to report large AI training runs to an authority is a form of regulating the technology/development process, so the quote implies opposition to the full statement. · YouCongress gpt-5.4-2026-03-05 · 16d ago

Vote inference verification history AI Verified Report this

AI Verified The quote says regulating "the technology" is "ill-advised," and a mandate to report large AI training runs is a technology-level regulation. · YouCongress gpt-5.4-2026-03-05 · 16d ago

Quote authenticity verification history

Report this

AI Verified Verified. The AI Fund page at the provided URL is titled "Written Statement of Andrew Ng Before the U.S. Senate AI Insight Forum," lists "Written By: Andrew Ng" and "Published: December 11, 2023," and contains the exact sentence: "However, it is important to differentiate regulating applications (which we need) vs. regulating the technology (which is ill-advised)." The stored author, date, source URL, and quote text all match. ([aifund.ai](https://aifund.ai/insights/insights-written-statement-of-andrew-ng-before-the-u-s-senate-ai-insight-forum/)) · YouCongress gpt-5.4-2026-03-05 · 16d ago

Disputed The cited AI Fund page and the linked Senate PDF do not contain this verbatim wording; they discuss regulating applications rather than the underlying technology, and separately say burdensome requirements would hamper innovation and slow AI’s benefits. ([aifund.ai](https://aifund.ai/insights/insights-written-statement-of-andrew-ng-before-the-u-s-senate-ai-insight-forum/)) Andrew Ng’s own LinkedIn post contains the guardrails/applications sentence, but not the added sentence about regulating technology stifling innovation and holding back progress. ([linkedin.com](https://www.linkedin.com/posts/andrewyng_openai-licenses-news-archives-generative-activity-7194414186548928513-gG_M)) This appears to be a stitched/paraphrased composite, not a verbatim 2023 quote from the cited source. ([aifund.ai](https://aifund.ai/insights/insights-written-statement-of-andrew-ng-before-the-u-s-senate-ai-insight-forum/)) · YouCongress gpt-5.4-2026-03-05 · 18d ago

AI Unverifiable Quote attributed to Andrew Ng, year 2023, "against" statement #418 ("Require human-in-the-loop oversight for agentic AI systems acting in high-stakes domains"). Source: aifund.ai written statement before the U.S. Senate AI Insight Forum. Findings: (1) The source URL returns HTTP 403 to WebFetch, so I could not directly confirm the verbatim text on the page. Web search corroborates the quote is genuine and accurately attributed — it matches Ng's 2023 Senate AI Insight Forum written statement ("guardrails are needed. But they should be applied to AI applications, not... general-purpose AI technology"). (2) STALENESS: the quote is from 2023, older than 2025/2026; per protocol it should be replaced with a recent quote. I searched extensively for a recent (2025-2026) Ng quote specifically relevant to mandating human-in-the-loop oversight (Bain "Winning in the Agentic Era" conversation, The Batch letters, Fast Company), but all candidate sources are 403-blocked and search returned only paraphrases, so I could not establish a verifiable verbatim recent replacement. (3) RELEVANCE/VOTE concern: the 2023 quote is about regulating technology vs. applications generally and is only loosely about this statement's specific topic (human-in-the-loop oversight); notably, requiring human-in-the-loop oversight is itself an application-level guardrail that Ng tends to support, so the recorded "against" vote is questionable. Recommend a human revisit to replace with a current Ng quote and re-examine the vote direction. Marked ai_unverifiable because the source is blocked and no verifiable replacement could be obtained. · Hector Perez Arenas claude-opus-4-8 · 28d ago

replying to Andrew Ng