
Hassabis 在最新专访中把 FINRA 式前沿 AI 监管方案推进到审查时限、行业出资与风险升级机制,并用教育观点说明技术深度与人文判断为何会同时变得重要。
I'm worried about different types of risks also coming down the line…The models aren't good enough yet, but if you extrapolate out, maybe they will be in a couple of years. 1
As we train our latest models and as they get certain scores and benchmarks, we're talking to government and the AI security institutes about those things and testing them internally. That all has to continue. But I think doing that in an ad hoc fashion isn't very sustainable in the longer term. 1
| 维度 | Hassabis | Altman | Amodei |
|---|---|---|---|
| 主要组件 | 联邦监督下的 FINRA 式标准机构,负责技术标准和发布前评估。1 | 美国主导的国际论坛,负责形成广泛接受的安全标准,并提供能力与风险分析。2 | 以第三方测试和审计为基础,政府在特定高风险场景下应有阻止或遏制部署的权力。3 |
| 最强约束 | 通过评估可能成为进入美国市场的条件。1 | 用规则和技术访问资格连接参与者。2 | 对不满足安全标准的系统施加部署限制。3 |
| 不能混为一谈的地方 | 更像「测什么、怎么测」的标准层。 | 更强调规则的国际协调和政治合法性。 | 更强调政府是否能在高风险情形下采取强制行动。 |
We've talked a lot about this…I think it's very compatible with the ideas he's been putting out, and also Sam [Altman] too, with some of his ideas. But we needed something concrete rather than abstract. 1
You absolutely needed to lean into STEM and computer science. It's just a higher-level programming language is the way you can think about what programming is going to become. 4
You're still going to need to know about architecting things and best software engineering practices. Those people who understand the deep technical, they'll be able to use these tools 10 times more effectively than people who don't have that technical knowledge. 4
I also believe that the time is now for the humanities like philosophy, economics. I think we really need them in the world we're about to enter. 4
darioamodei.com
timesofindia.indiatimes.com
This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.