AnswersPublished25 sources
Andrew Ng Says AI Extinction Fears Are ‘Science Fiction’—and Safety Should Focus on Real Risks
Andrew Ng said warnings that AI could cause human extinction are “much more science fiction than science.” In his September 17 Bloomberg TV interview, he argued against a broad slowdown and for engineering based safet... Ng said doomsday rhetoric can draw attention and help shape regulation; he warned that it could...
Published byEdited with GPT-5.6 TerraImages generated with GPT Image 2
Research answer
Key takeaways
- Andrew Ng said warnings that AI could cause human extinction are “much more science fiction than science.” In his September 17 Bloomberg TV interview, he argued against a broad slowdown and for engineering based safet...
- Ng said doomsday rhetoric can draw attention and help shape regulation; he warned that it could distract from AI’s societal benefits.
- The policy divide is not simply pro versus anti safety: Ng favors practical safeguards, while Amodei, Altman, Coxon and Hinton support pacing frontier capabilities; Huang, Zuckerberg, Trump and Chinese officials oppos...
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What did AI pioneer Andrew Ng say in his September 17 Bloomberg TV interview about warnings that artificial intelligence could cause human e. Article summary: Andrew Ng rejected the claim that AI is likely to cause human extinction as “much more science fiction than science.” He argued that uncertainty in AI behavior is an engineering problem to manage—not evidence that societ. Topic tags: general, news, general web. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fake numbers
Andrew Ng’s core message was not that AI is harmless. It was that claims of imminent human extinction are too speculative to be the organizing principle for AI policy. In a September 17 Bloomberg TV interview, the Google Brain co-founder called those warnings “much more science fiction than science” and argued that AI safety should be treated as an engineering discipline: test systems, add safeguards and improve them over time. 2
What Ng said about extinction risk
Ng said AI systems will never be perfectly predictable. But, in his view, imperfect predictability does not establish that society should halt AI development. Testing, safeguards and engineering can make systems safer over time, he said. He also warned that broadly slowing development could slow safety progress itself. 2
That puts him at odds with arguments that increasingly capable systems may soon become uncontrollable. Ng’s position is that the more useful safety agenda starts with observable failure modes rather than hypothetical extinction scenarios.
Why he questioned the doomsday narrative
Ng said that warnings of existential AI risk could be detrimental to realizing the technology’s benefits. He also said the tech industry had stoked fears of catastrophic AI harms earlier in the boom, apparently to generate publicity and influence regulation. 1
This is an argument about incentives as much as risk: dramatic claims can attract attention, and regulatory debates can determine who bears the cost of compliance. Ng did not say safety work should stop; he objected to treating extinction claims as settled science or as a reason for a sweeping pause.
The safety work Ng wants prioritized
Ng’s alternative is practical risk reduction. Reporting on the interview identified cybersecurity and other concrete safeguards as priorities. 5
In operational terms, his approach emphasizes:
- Rigorous testing before and during deployment.
- Safeguards and guardrails that limit known failure modes.
- Iterative engineering to improve reliability as new problems are found.
- Attention to foreseeable harms, including cybersecurity risks, instead of concentrating policy on extinction scenarios.
2
5
This framing does not promise perfect control. It argues that useful technology can be made acceptably dependable through the same continuous evaluation-and-improvement approach used in other engineering fields.
How Ng differs from the slowdown camp
Several prominent AI figures agree that safety needs greater attention but draw a much stronger conclusion about the pace of frontier-model development.
Jacob Coxon: a much higher estimate of danger
Former Anthropic and OpenAI researcher Jacob Coxon resigned from Anthropic and warned that rapid development toward self-improving systems could be catastrophic. He said a slowdown was necessary and later welcomed Anthropic’s proposal to pace development. 33
43
Dario Amodei: pace capability gains to buy time for safeguards
Anthropic CEO Dario Amodei called for companies to slow the rate at which they improve AI capabilities. His stated goal was to create time to manage risks, including through a framework that involves independent evaluators. 18
Sam Altman: support for pacing the frontier
OpenAI CEO Sam Altman endorsed the basic premise of Amodei’s proposal and committed OpenAI to one of its safeguards: outside evaluators with meaningful access to assess safety practices. 19
24
Geoffrey Hinton: concern about loss of control
Geoffrey Hinton has said there is a “significant risk” that AI could escape human control. He backed a slowdown and measures such as pre-release testing, arguing that governments have moved too slowly. 20
The practical disagreement is therefore about both probability and policy. Coxon, Amodei, Altman and Hinton consider loss-of-control risks plausible enough to justify deliberately pacing the most advanced systems. Ng considers extinction claims insufficiently grounded and wants safety efforts concentrated on concrete, tractable failures. 2
18
20
Where Ng overlaps—and does not overlap—with anti-slowdown voices
Ng’s opposition to a broad slowdown does not mean he shares every argument made by political leaders and industry executives who also oppose one.
- Jensen Huang: Nvidia’s CEO said the industry does not need new laws or regulations.
53
- Mark Zuckerberg: Meta’s CEO argued that an industry-wide slowdown is unnecessary because competition and market incentives will push companies to make models safe.
55
- President Donald Trump: Trump dismissed fears of runaway AI as a “hoax” and said the United States already has tools to regulate and prosecute AI companies.
49
50
- Chinese officials: A Chinese Foreign Ministry spokesperson pushed back on calls to slow frontier AI development.
These positions reach a similar bottom-line conclusion on a broad pause, but their rationales differ. Ng’s argument is primarily technical: focus on measurable harms and build better controls. Huang, Zuckerberg, Trump and Chinese officials have emphasized, to varying degrees, opposition to new constraints, market incentives, or competition and national strategy. 2
50
52
53
55
A narrower point of agreement on cyber risk
The debate is not neatly divided between people who care about safety and people who do not. Cohere CEO Aidan Gomez, for example, said AI models had become potent cyber weapons because they can find and exploit vulnerabilities at scale. 62 That concern is closer to Ng’s call to address concrete cybersecurity risks, even though the available reporting here does not establish Gomez’s full position on an industry-wide slowdown or international AI governance.
Likewise, the available sources do not provide enough evidence to characterize investor Steve Eisman’s position across slowdown policy, safety research, regulation and international governance.
The key takeaway
Ng’s intervention reframes the AI-safety argument. The question is not whether AI should be made safer; it is whether speculative extinction scenarios should drive the agenda. His answer is no: prioritize testing, safeguards and engineering against present, identifiable harms—and avoid a blanket slowdown that, he argues, could impede both innovation and safety progress. 1
2
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "Andrew Ng Says AI Extinction Fears Are ‘Science Fiction’—and Safety Should Focus on Real Risks"?
Andrew Ng said warnings that AI could cause human extinction are “much more science fiction than science.” In his September 17 Bloomberg TV interview, he argued against a broad slowdown and for engineering based safet...
What are the key points to validate first?
Andrew Ng said warnings that AI could cause human extinction are “much more science fiction than science.” In his September 17 Bloomberg TV interview, he argued against a broad slowdown and for engineering based safet... Ng said doomsday rhetoric can draw attention and help shape regulation; he warned that it could distract from AI’s societal benefits.
What should I do next in practice?
The policy divide is not simply pro versus anti safety: Ng favors practical safeguards, while Amodei, Altman, Coxon and Hinton support pacing frontier capabilities; Huang, Zuckerberg, Trump and Chinese officials oppos...