Ask HN: Is anyone else bearish on OpenAI?

Before ChatGPT, to find the answer to things like "how do I set up Gunicorn to run as a daemon that restarts when it fails" I would have to endure hours of googling, snarky stack-overflow comments that I shouldn't do that, etc., but as a solopreneur without access to a more senior engineer to ask, it's been fantastic. I've been quite skeptical of machine learning/AI claims but I feel like I'm experiencing a genuine case of a technology that's proving to be so much more useful than I had imagined.

Crypto, after more than a decade, has been useful only to criminals and scammers.

LLMs aren’t AGI. They’re far it. But they have massive uses for reasoning on available context.

I’ll give you an example. I’m trying to set up some bulk monitoring for api across 200k jvm s. The api documents are horribly out of date. But I get the raw uri on the monitoring tools.

I can just get these uri, send them into chatgpt and ask for a swagger spec - along with a regular expression to match the uri to the swagger api. It figures out the path and query params from absolute paths.

Sure I could try to figure out how to do this programmatically using some graph or tree based algorithm. But chatgpt basically made it possible with a dump Python script.

Of course I may still need a person to fill in these. But just getting a swagger spec done for a thousands of services in an afternoon was awesome.

Imagine if they kept doing that despite having your credit card information because you paid for Pro, which more or less proves you're an adult who deserves the presumption of innocence/good-actordom.

Lastly, imagine that you do this for all users despite the fact that it is known to reduce the intelligence of your output.

(I'm bullish on self-run models)

Sometimes you need to go back and forward a bit, I tell ChatGPT it’s wrong and the error message and then it spits out the correct result, sometimes I need an algorithm tweaked because it has assumed a wrong constraint on the problem, again just explain it clearly and unambiguously and it will make corrections, there’s only been maybe 1/40 problems I couldn’t get a correct answer on after (sometimes a lot!) of back and forwards.

I am not looking for a perfect oracle, I am looking for something to write 80% of the code and then I’ll fix it up. It’s still way faster this way, especially in domains I don’t know, E.g. I just learned CUDA with ChatGPTs help.

It’s not perfect, and neither am I, but it doesn’t have to be perfect to be useful, you can get to millions in revenue through 80% solutions