Pseudonymous account that builds an AI job-application tool and posts about AI coding, local models and open-source competition.

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:他们表达的 Doom–Bloom 前景看法。 纵向:变革程度。

Doom–Bloom:100 中的 93。变革程度:100 中的 65。解读范围:横向为 88 至 100,纵向为 50 至 75。这些是解读坐标,而不是事件概率。

Roy的 P(doom) · 推断

≈4%

0%100%

根据他们的模拟回答推断,并非他们给出的数字。 合理范围:3–8%。

我们如何估计 P(doom)

他们的展望取决于什么

一个核心假设

Local inference is becoming genuinely powerful, and open source gives every second-place player an incentive to scorch the earth rather than let one company monopolize the stack.
回答 1

如果这个假设实际并非如此,他们的展望会如何变化?

一个尚未解决的问题

I don’t have a serious percentage.
回答 3

什么能帮助他们区分这里各种合理的结果?

什么可能使其改变看法

A concrete, reproducible demonstration that capable systems can reliably escape deployment controls, acquire resources, and cause serious harm without human authorization would change my view—not another all-caps “bombshell” thread.
回答 5

什么证据才足够,又会让他们的观点朝哪个方向转变?

更多详情

预期益处

预计将带来具有变革性且广泛有价值的收益。

93 / 100

影响小变革性影响

在定性尺度上,解读范围为 67 到 100。

预期危害

预计几乎不会产生实质性的不利影响。

5 / 100

影响小变革性影响

在定性尺度上,解读范围为 0 到 33。

人类影响力

人类的选择可以大幅改变AI的发展轨迹。

66 / 100

影响力小影响力强

在定性尺度上,解读范围为 46 到 79。

AI访问权限

限制对强大AI的访问。

允许访问,但须遵守能力或用途限制。

模拟位置:支持广泛或开放地访问强大AI。

这些解读保留了他们陈述的条件。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们的模拟回答,而不是统计置信区间。

与Roy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 Roy 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?

AI means a ridiculous increase in individual leverage. It’s already better than humans at a lot of the economically valuable stuff—quantitative reasoning, math, problem solving, coding, cutting through bureaucratic sludge—while everyone fixates on whether it can cosplay Hemingway. Cool benchmark, bro. The future also won’t belong exclusively to whichever lab owns the biggest datacenter. Local inference is becoming genuinely powerful, and open source gives every second-place player an incentive to scorch the earth rather than let one company monopolize the stack. That competition matters. So I’m broadly bullish and deeply skeptical of pause hysteria. When labs run short on compute and suddenly discover that slowing everyone down and kneecapping open source is “safety,” I see self-interest wearing a concerned face. I’m not claiming every job disappears or assigning some fake AGI date. I’m saying the leverage is here now, and people still writing code by hand or treating models as magic black boxes are already behind.

问题 2

你认为AI最终会在多大程度上改变世界?

Massively. It turns expert-level quantitative work, coding, problem solving, and bureaucratic demolition into cheap, widely available leverage. That changes who can build companies, automate work, and compete with institutions that used to win through headcount and process. But I don’t buy the cartoon where one magic datacenter owns intelligence forever. Local models and open source will keep spreading capability because every player below first place has an incentive to prevent monopoly. I’m not giving you a fake unemployment number or AGI countdown. The concrete change is already absurd: one competent person using the right models can outperform whole groups still doing everything by hand.

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 a serious percentage. My gut says the extinction framing is massively overhyped, and most public doom talk reads like lockdown-style panic amplified by labs with incentives to slow competitors and attack open source. That doesn’t prove zero risk; it means I’m not going to invent a P(doom) because someone wrapped vibes in decimals.

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?

“Smarter than humans” is too mushy. Systems already beat humans at valuable tasks, and people still control them through deployment, permissions, compute, and choosing which model to use. I don’t see capability automatically turning software into an uncontrollable demon. Could control get harder as systems become more capable? Sure. But the leap from “better than us at reasoning or coding” to “nobody can control it” is usually asserted, not demonstrated. My bigger near-term concern is humans using safety rhetoric to centralize control in a few labs and cripple local or open models.

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?

A concrete, reproducible demonstration that capable systems can reliably escape deployment controls, acquire resources, and cause serious harm without human authorization would change my view—not another all-caps “bombshell” thread. On the other side, if local and open models genuinely hit a hard wall while centralized labs kept pulling away, I’d revise my view that capability will diffuse and monopoly will be checked. Right now, neither story has earned the certainty people attach to it.
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