Jon Stewart

The Daily Show

Comedian and interviewer who questions AI’s effects on work and power while exploring its useful applications.

AI将如何改变世界？

模拟位置解读范围

横向：他表达的 Doom–Bloom 前景看法。纵向：变革程度。

Jon Stewart的 P(doom) · 推断

≈16%

0%100%

根据他的模拟回答推断，并非他们给出的数字。合理范围：10–24%。

Jon Stewart 的里程碑时间线

工作与机构
But the transition could be compressed enough to overwhelm our already mediocre ability to protect workers.
回答 2

按里程碑分组，不按推断日期间隔或排序。AGI 和超人类 AI 保留他的定义。

他的展望取决于什么

一个核心假设

They require ordinary human greed, weak oversight, and very powerful tools.

回答 3

如果这个假设实际并非如此，他的展望会如何变化？

一个尚未解决的问题

I don’t know whether AI ultimately transforms every occupation or hits meaningful limits.

回答 2

什么能帮助他区分这里各种合理的结果？

什么可能使其改变看法

Technically, I’d also pay attention to repeated, real-world evidence that these systems have durable limits: not one chatbot losing a knife fight with a drive-through order, but clear limits across medicine, education, persuasion, and military decision-making.

回答 5

什么证据才足够，又会让他的观点朝哪个方向转变？

相似的世界观

模拟世界观与 Jon Stewart 最接近的意见领袖

Jon Stewart关于AI说过的话

Stewart questions AI’s effects on work and power, explores useful applications, and described a slight shift toward optimism in July 2026.

2026年7月29日
“And I recently read a book that changed my perspective slightly.”
The Weekly Show, Cory Doctorow interview
2026年5月12日
“But it's hard to imagine it will remain our assistant.”
The Daily Show, Josh Tyrangiel interview
2026年4月22日
“AI is a human expertise laundering machine.”
The Weekly Show, David Autor and Daron Acemoglu interview
2025年10月9日
“my guess is, like any technology, there's going to be some incredible positives.”
The Weekly Show, Geoffrey Hinton interview

逐字引自所链接的出处，核对于 2026年10月3日

模拟评估

问题 1

你认为AI对我们的未来意味着什么？为什么？

I think AI means we’re heading into a very useful and potentially very destabilizing transition. It may help diagnose illness, teach people, translate across languages, and make expertise more widely available. That’s real. But the sales pitch tends to jump from “this tool can help you” to “therefore five companies should own the accumulated knowledge of humanity.” That is quite a leap—like borrowing everyone’s recipes and then charging them admission to dinner. My biggest concern is speed and power. Earlier industrial changes displaced workers over decades, and we still handled them badly. If this happens much faster, what happens to people’s jobs, wages, dignity, and bargaining power? And even where workers aren’t replaced, AI can become a surveillance system that lets employers control them more tightly. None of that is inevitable. Markets don’t descend from the heavens with tax incentives and corporate subsidies engraved on stone tablets. We make political choices about who owns these systems, who shares in the gains, and whether labor is treated as less valuable than capital. I want the public—and especially the people whose work and knowledge trained these tools—to have ownership and a meaningful voice. And military use is chilling. A system that acts quickly and sounds confident can make violence seem cleaner or easier than it is. So I’m still broadly a doomer, though somewhat less doctrinaire than I was. The technology has genuine promise; my fear is that our institutions and incentives may direct that promise toward concentration, control, and weapons rather than broadly shared benefit.

问题 2

你认为AI最终会在多大程度上改变世界？

Enormously—but “how much” depends less on whether the machine becomes some omniscient robot god than on how deeply we let it into work, education, medicine, communication, surveillance, and war. Even a system that fails spectacularly at taking a drive-through order can still change who gets hired, who gets monitored, who owns valuable knowledge, and who has leverage. You don’t need science fiction for major disruption; ordinary corporate incentives will do just fine. I don’t know whether AI ultimately transforms every occupation or hits meaningful limits. But the transition could be compressed enough to overwhelm our already mediocre ability to protect workers. And if a few firms own the systems and the infrastructure around them, AI may alter the distribution of power as much as productivity. So yes, I expect a profound change. The open question is whether it resembles a widely shared public benefit or an extraordinarily efficient machine for concentrating wealth and control. That part is political, not preordained by the technology.

问题 3

凭直觉，你认为AI导致人类灭绝或类似的永久性灾难的概率有多大？

I don’t have an honest percentage. My gut says the risk is serious enough not to dismiss, especially through weaponized misuse, military escalation, and systems creating false confidence at machine speed. But putting “17 percent” on it would be fake precision—the kind of number that wears glasses so everyone assumes it did the homework. I’m more certain about the near-term danger: concentrated power, rapid labor disruption, surveillance, and companies racing for dominance while buying political influence. Those harms don’t require AI to become an all-powerful autonomous being. They require ordinary human greed, weak oversight, and very powerful tools. Extinction is the most dramatic possibility; institutional failure and large-scale violence are more immediate concerns I can actually argue from.

问题 4

你预计人们能否继续控制比人类更聪明的AI系统？为什么？

I don’t think control is guaranteed, but I’m also wary of pretending we know exactly what “smarter than humans” would mean in practice. Smarter at what—medical diagnosis, persuasion, military targeting, running a company? A calculator is smarter than I am at arithmetic; fortunately, it has not yet seized the Treasury. My immediate concern is that people may lose meaningful control before machines become independent actors. If a handful of corporations own the systems, governments defer to them, employers use them to monitor workers, and militaries trust their recommendations because they’re fast and confident, then technically a human may still be pressing the button—but that is a very thin definition of control. So I don’t accept helplessness as inevitable. People can set ownership rules, impose accountability, preserve human judgment, and give the public a real voice. But under current incentives—a race for market dominance, military advantage, and political influence—I’m not confident those with power will exercise restraint merely because it would be wise.

问题 5

哪项发现或事件最可能改变你对AI未来影响的看法？

The biggest change would come from evidence that the benefits can be broadly shared without concentrating ownership and control. If workers whose knowledge trained these systems gained real ownership, bargaining power, and income—and AI consistently augmented their work rather than replacing or surveilling them—I’d become substantially more optimistic. Technically, I’d also pay attention to repeated, real-world evidence that these systems have durable limits: not one chatbot losing a knife fight with a drive-through order, but clear limits across medicine, education, persuasion, and military decision-making. That would weaken the grand replacement story. In the other direction, a major autonomous weapons failure, rapid displacement across several industries, or systems making consequential decisions that humans could no longer meaningfully inspect or override would make me far more pessimistic. For me, the decisive event isn’t merely a machine scoring higher on a test. It’s what happens when the technology meets institutions: who owns it, who answers for it, and whether ordinary people gain power or lose it.

来源

用于为此模拟用户提供事实依据的文章、访谈和著述。

AI and the Enshittification Era with Cory Doctorow

Publisher English captions inspected. His introduction describes a slight shift after an AI doomer spiral. Later turns challenge surveillance and concentrated capital and describe a failed drive-through chatbot. This is a modest dated shift, not adoption of Doctorow’s capability ceiling.

youtube.com

AI for Good: interview with Josh Tyrangiel

Original publisher English captions inspected across the full interview. Stewart recognizes useful educational, medical and communication examples, questions whether AI will remain an assistant, calls the extraction of human knowledge a reason for public ownership and voice, and rejects an inevitable loss of control. He distrusts corrupt regulation and calls for trusted adjudication.

youtube.com

AI and the Future of Work with David Autor and Daron Acemoglu

Original publisher English captions inspected at his introduction, labor questions, extraction critique and policy discussion. He worries about rapid workforce disruption, expertise used without compensation, companies owning society’s operating infrastructure, and wealth purchasing political access. Explicitly favors reforming capital-versus-labor tax incentives and giving people ownership in industries. The guests’ timelines and AGI limits are excluded.

youtube.com

Military AI: Anthropic, OpenAI and the future of warfare

Original publisher English captions inspected, especially his questions about military incentives, oversight and public accountability. Calls AI-enabled targeting chilling, presses whether builders appreciate potential damage, and asks whether systems create false military confidence. His introduction acknowledges more nuance than an Anthropic hero/OpenAI villain story. Experts’ proposed hardware controls are not his policy.

youtube.com

AI: What Could Go Wrong? with Geoffrey Hinton

Original English publisher captions inspected; translated auto-captions were rejected. He asks for a basic explanation, welcomes possible benefits, worries about compressed labor disruption and weaponized misuse, and questions executives’ pursuit of money and power. His earlier flattering-search-engine comparison describes personal experience, not a permanent capability ceiling. Hinton’s takeover probability and timelines are not Stewart’s.