It’s probably almost cruelly difficult to find someone with up to date machine learning knowledge who also violently hates slop as much as they ought to.

Edit: obviously I mean hating AI the buzzword, AI the machine god, AI the magic bean stock goes up, AI the cure to critical thinking, AI will change everything, AI = LLM = photo generator. Not hiring someone who abhors the concept of machine learning. I think most people here got what I was thinking of I guess

And that’s why you don’t have a business.

If you want a machine learning model that actually works, you need to feed it good data. If you hire an AI slop enthusiast to feed the model mass junk data, you will have a bad time.

And when I say "actually works", I mean works 100% of the time. If you are using the machine learning model for something like vision based part inspection/detection, a 99% success rate will be unacceptably low.

Depends on what you need doing. Machine learning hasn't progressed as much as you might think and there's plenty of us willing to use it where it's actually valuable.

The problem for many of us is generative LLMs and the forced nature of the rollout with zero regard for impact. A lot of the "progress" on them is not on the underlying machine learning parts as much these days. They focus more on the data, adding more parameters, or pre/post processing prompts/output.

Hating AI and hating slop are different. They want someone who hates both.

If you want a machine learning model that actually works, you need to feed it good data. If you hire an AI slop enthusiast to feed the model mass junk data, you will have a bad time.

You understand the strawmanning and logical fallacy here right? A person can appreciate novel technologies from afar, understand when/if theyre appropriate for their job, and then do their unrelated work just fine without a personality purity test

This right here. AI/LLMs by itself is not a problem. It’s merely a tool that can (and often is) abused for whatever purpose. The problem is that it’s being forced on us, used without consent, and unable to be disabled.

Yesterday I was having a problem in my code that I just couldn’t figure out on. So I asked AI, and in a few minutes it pinpointed a configuration I misapplied, and now my code works. It wasn’t documented not to do it the way I was doing it either. And to be honest, I don’t know if I’d have ever found the issue on my own; at least not in a timely manner.

You understand the strawmanning and logical fallacy

No one gives a fuck that you joined your high school debate club.

OP specifically mentioned trust. There's a big difference between knowing that it's possible for a person to separate their enthusiasm for AI from their ML day job, and trusting them to actually do it.

I'd look for someone with a nuanced view as the extreme ends of almost any spectrum rarely arrived there through wisdom or rationality.

The industry I work in (game dev) hates Gen ai but loves ML

It’s merely a tool that can (and often is) abused for whatever purpose

Louder for the people in the back. Level-headed and reasonable views are very much appreciated, and I’m finding are often in short supply.

It's possible to disagree without insults and rudeness. That was completely uncalled for.

also the copyright infringement thing that makes it a huge risk to integrate

You’re right. That and the environmental impacts and economic losses.

How is machine learning generally used there?

Even Ed Zitron has mellowed out in the last year with the rationalist opinion of "it's a probabilistic model being asked to do deterministic things," and "it makes the easy things easier and the harder things harder."

The overwheling majority of "anti AI" comments I see are either 1) anti LLM, B) anti hyperscaler business / marketing practices, C) anti data center, or usually D) any combination of the above. Very little about the objective, architectural functions of language models.

Your work sounds interesting. any article or wiki you recommend about the subject?