HACKER Q&A
📣 data_maan

Is the LLM craze real? Or are companies just hyping it to fish for CVs?


We all read the news about company X,Y,Z (e.g. X=Microsoft, Y=Google, Z=OpenAI) spending a lot of $$$ on building hyped-up LLMs and hiring talent for that (to the detriment of other CS areas, where there is a happy cull going on :(). This obviously drives application numbers.

In particular, there are jobs that do not require a mastery of a very large skillset, like prompt engineer (sorry, if I'm offending any prompt engineers out there), that yet offer a startking salary of 280k: https://jobs.lever.co/Anthropic/e3cde481-d446-460f-b576-93cab67bd1ed

This job has been online at Anthropic since more than 2 months now and I personally know someone who is a final-year PhD student in machine learning at a top-5 university, with publications at top conferences which were precisely on prompt engineering. You'd think that you can't get a much better candidate - yet he didn't get the job; he actually wasn't even invited to do an online quiz or a first interview.

These anecdotes, the hype that companies push out as well as the fact that these jobs stay out there forever while the market is down for other CS jobs smell to me that companies are not really looking for talent, just for CVs. How do you see this?

  👤 genezeta Accepted Answer ✓
Did your acquaintance provide "a specific prompt engineering project on LLMs that you’re proud of in your application"? Was that project interesting in a particularly innovative way?

I mean, I don't really know but an alternative explanation to "CV fishing" can be "idea fishing". You have a startup with something they call a "product", their generic assistant Claude. They've invested in feeding and growing a LLM. But while they claim this assistant can be integrated "anywhere" to solve "multiple tasks", that is all pretty generic. Which means -maybe; again I don't know- they lack actual practical scenarios where the application of a LLM will provide direct and visible benefits. That is, they got something they think can be useful but they still need to find how to make it so, how to provide interesting use cases. And so, they would be interested in getting all these small, experimental ideas to read and think about.