Hacker News Reader

Comment by tencentshill | Hacker News Reader

Why did they bother hiring someone to write at all? I certainly wouldn't invest $1T into a company with so little confidence in their product. When the chips are down and they need to write something important, they DON'T use AI?

View on HN

Obviously everyone at Anthropic uses a ton of AI in everything. The point here is that they clearly use it well, and that they use it as a tool to augment and enhance what they do rather than as a replacement for effort. The blog posts don't read like the output of a prompt because they're probably the output of dozens of prompts, rewritten and edited, by someone who's using AI to actually write something worth reading.

Anyone can having blog posts as good as this by using AI. Just not by zero-shotting it with a 2 line prompt. It still takes a lot of work.

But if you need to pay brilliant people 500k base salary to use AI well then what is a smaller company supposed to do?

There aren't going to be enough experts to go around. They'll start demanding millions in salary and all of a sudden the advantage is gone.

I don't think this is an AI specific problem. Small companies have never been able to compete on salary with Google, Netflix, etc. They need to focus on growing talent instead of buying it. And that talent will leave eventually, so there also needs to be a focus on culture, and a pipeline to replace people. AI won't change that.

Do you also demand that McDonald’s employees only eat McDonald’s?

They use their product to do what it’s best at, and use humans where it isn’t yet good enough.

That's just picking a fight. Whatever we think about how amazing or terrible AI is, it's pretty reasonable to understand it is quite literally the weighted average of humanity's output.

Also reasonable: it's possible to hire a world-class ___ to do that job better than AI.

It's not a weighted average though. I see this repeated all the time. Between synthetic data, custom produced training data by experts, post-training regimes etc the models are far more diverged from that baseline at this point, at this point it's mostly a right shifted curve.

I hesitated to write that line because I don't know the math. So thanks for clarifying. Main point is the output is along some distribution and I'd say AI-evangelist themselves see "good enough" across everything and anything as a feature.

Their investors expect true better-than-human intelligence. There's only about 30% of the entire US economy invested in that expectation right now.

Reply on news.ycombinator.com