Waymo says there's no AI shortcut to self-driving

4 pointsposted 10 hours ago
by samizdis

3 Comments

nialse

9 hours ago

Definitely AI slop. It is interesting that in such a short amount of time the brain has learnt a new aversive pattern, and refuses to engage with it. I still have no clue what Waymo said. The brain just skipped the attention part when detecting AI slop. Curious.

dofm

6 hours ago

The Axios article? Probably not slop, at least not quite in the way you might think.

This is how Axios writes — it’s their house style for these forms of articles — and have done for several years before LLMs. This style (Axios call it “Smart Brevity” and they made a fuss about it to the point of being lampooned) was fully in place in 2017 when they were founded. Search back a bit and you will find it.

As a (maddening) style it has been quite influential; it may have influenced OpenAI products’ writing style. Axios being sort of politically centrist and tech focussed must surely have been appealing in the Valley.

It is actually patented. Axios built a system that could take an existing journalistic article and guide the writer or editor into restructuring it in this format.

They wrote a book about it:

https://www.axios.com/smart-brevity

And built a business selling the system:

https://www.smartbrevity.com/

Back in 2023 they accused the NYT of nicking the idea.

I assume it is LLM-guided now, but it’s not correct to assume that it’s a simple LLM text output: they put way more into this and it definitely happened five years before ChatGPT.

It is deeply annoying but you’re not looking at ordinary lazy AI slop in the conventional sense. It’s much more deliberate.