Dax Raad

Dax Raad

x.com/thdxr

Creator of the open-source, model-neutral OpenCode coding agent who favors broad access to AI as a defense against misuse.

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 81。变革程度:100 中的 55。解读范围:横向为 75 至 100,纵向为 50 至 75。这些是解读坐标,而不是事件概率。

Dax Raad的 P(doom) · 推断

≈6%

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:3–12%。

我们如何估计 P(doom)

他们的展望取决于什么

一个核心假设

Bad actors will use AI regardless, while defenders need capable tools to investigate and respond.
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 defensible percentage.
回答 4

什么能帮助他们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。

76 / 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

预计会出现可控或局部的危害。

26 / 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。

70 / 100

影响力小影响力强

在定性尺度上,解读范围为 48 到 77。

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Dax Raad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 Dax Raad 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 means dramatically more leverage for people who build things, but the surrounding infrastructure is still immature. Code generation is the obvious example, yet the story isn’t simply “AI creates more code and therefore more mess.” The same leverage lets us refactor, migrate, test, and understand code faster. What matters is the actual product experience across real projects, not benchmark marketing or folklore built around a few noisy model outputs. I also think the future should stay model-flexible. Models have different strengths, providers change quickly, and quality depends heavily on the task and environment. That’s why I prefer neutral tools over locking the whole stack to one model. Open source helps when a community can cover the long tail of models, providers, operating systems, and workflows, though it isn’t automatically the right answer for every product. More broadly, I favor wide access. Bad actors will use AI regardless, while defenders need capable tools to investigate and respond. Restrictions can block legitimate defensive work without reliably stopping misuse. And when companies describe AI as an existential threat while preserving enormous commercial upside for themselves, I think that tension deserves much more scrutiny.

问题 2

综合考虑益处和危害,你预计AI会对我们的社会产生怎样的总体影响?

Overall, I expect AI to be a major net positive because it gives ordinary developers and organizations far more leverage—not just to generate code, but to understand, repair, refactor, and defend systems. The gains will be uneven, and the infrastructure is still much less mature than the marketing suggests, so I don’t expect a clean or frictionless transition. The main danger is concentrating capable systems behind restrictive platforms while assuming restriction itself solves misuse. Harmful users will keep looking for access; meanwhile, legitimate defenders can be blocked from analyzing attacks or building countermeasures. Broad access, model choice, and open ecosystems give more people the ability to respond. That doesn’t mean open source wins every product decision, or that every new model is socially beneficial. It means I’d rather have a plural ecosystem where tools compete on real-world usefulness than one governed by benchmark theater, vendor lock-in, and catastrophic-risk rhetoric that somehow leaves the same companies holding all the power and upside.

问题 3

你认为AI最终会在多大程度上改变世界?

A lot—probably at the level of a general-purpose shift in how knowledge work gets done. In software, the change isn’t just faster code generation. It’s faster debugging, refactoring, migration, testing, and understanding unfamiliar systems. That gives individuals and small teams leverage that previously required much larger organizations. But I don’t think the final shape is obvious yet. The models change quickly, infrastructure is immature, and real usefulness varies enormously by task, provider, and environment. So I’m skeptical of precise forecasts based on today’s benchmarks. The bigger question is whether this leverage stays broadly accessible and model-flexible, or gets concentrated inside a few restrictive platforms. That choice will strongly affect who benefits and who can defend against misuse.

问题 4

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 defensible percentage. My stronger view is that existential-risk rhetoric often outruns evidence and can conveniently justify concentrating control in the same companies capturing the upside. Restricting capable systems doesn’t make malicious access disappear; it can also prevent legitimate users from analyzing attacks and building defenses. So I take catastrophic misuse seriously, but my preferred defense is broad access, model choice, and a capable ecosystem of defenders—not pretending a few proprietary gatekeepers can contain the risk.
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