Jason Kneen

Jason Kneen

x.com/jasonkneen

UK app developer who builds open-source AI agent tools and posts about how prompts and agents will change the way software is made.

AI将如何改变世界?

文明层面的变革渐进式变化DoomBloom
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。

Doom–Bloom:100 中的 75。变革程度:100 中的 54。解读范围:横向为 70 至 80,纵向为 49 至 76。这些是解读坐标,而不是事件概率。

Jason Kneen的 P(doom) · 推断

≈5%

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:低于 24%。

我们如何估计 P(doom)

他的展望取决于什么

一个核心假设

The blockers are still time, cost and reliability.
回答 1

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

“Smarter than humans” covers a lot, and I haven’t worked out whether meaningful control remains possible at that point.
回答 4

什么能帮助他区分这里各种合理的结果?

更多详情

预期益处

预计将带来显著益处,但受到重要条件或分配方面的限制。

72 / 100

影响小变革性影响

在定性尺度上,解读范围为 67 到 100。

人类影响力

根据你的回答得出的暂定估计;较宽的范围表示其他合理解读。

51 / 100

影响力小影响力强

在定性尺度上,解读范围为 0 到 100。

这些解读保留了他陈述的条件。益处和危害都可能很大。这些范围描述的是我们如何解读他的模拟回答,而不是统计置信区间。

与Jason Kneen相比,你的立场在哪里?
用大约3分钟描绘你自己的AI世界观,然后进行比较

相似的世界观

模拟世界观与 Jason Kneen 最接近的意见领袖

来源

用于为此模拟用户提供事实依据的文章、访谈和著述。

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 changes how we build software and media more than it simply “takes everyone’s jobs.” Today’s models are astonishing, but they’re also unreliable and need plenty of steering — I’ve watched them misunderstand very simple things, lol. The interesting bit is the workflow shift: prompts, or whatever we eventually call them, could become the new App Store content. Instead of patching software, we update the spec or prompt and rebuild an app from small, self-contained pieces. Media heads toward on-demand generation too, with rights holders, platforms and creators sharing revenue. The blockers are still time, cost and reliability. That’s why open-weight models and free or cheap inference matter — builders need access, not just another expensive plan from a big lab. So I’m optimistic about augmentation and creativity, but the hype is a LONG way ahead of what these systems can consistently do.

问题 2

你认为AI最终会在多大程度上改变世界?

Massively — especially in how software and media are created and distributed. I expect apps to become generated, modular things rebuilt from prompts or specs rather than manually patched, while entertainment becomes increasingly generated on demand. But “massive change” doesn’t mean today’s models are about to replace everyone. They’re powerful, unreliable, and still need steering. The long-term shift could be enormous; the current hype often pretends we’ve already arrived, and we really haven’t.

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 honestly don’t have a percentage. I haven’t formed a firm position on extinction risk, and plucking out a P(doom) number would be fake precision. My work is focused on what current models can actually do—and right now they’re powerful, useful, and often hilariously unreliable. That doesn’t answer the long-term catastrophe question, but I’m not going to invent certainty where I don’t have it.

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 have a firm view on that. “Smarter than humans” covers a lot, and I haven’t worked out whether meaningful control remains possible at that point. My experience is with current systems: powerful, unreliable and heavily dependent on steering. Extrapolating from those tools to superhuman AI would mostly be guesswork.
你的立场在哪里?
回答几个简单问题,探索你自己的AI世界观。
描绘你自己的世界观

你的立场在哪里?

描绘我的世界观