Post

increasingly believing much of the communication gap between LLM Enthusiasts and LLM Skeptics on this website boils down to “are you someone who has largely experienced a *reasonable* amount of job security in tech, or are you someone who has not experienced job security and likely aren’t in tech”

I think this is super real, and I want the conversation to move here. The longer we stay in the "is it useful" debate, the longer we defer taking action to protect ourselves in a changed world.

I am pretty concerned about a world where there's only 2-3 companies that can run these models. I have been spending the last few days idly musing about a coop that sets up hardware and runs the open models.

Regarding the employment future around software, it's hard to make predictions right now without knowing where the model capabilities land. I barely know how to think about hiring right now.

My most optimistic take is that IT departments in non-software-oriented companies might be able to expand their scope to creating and maintaining inhouse software. Based on the level of expertise still required in the current generation of models, I think that's a reasonable idea

This would essentially displace the saas industry and distribute the jobs outwards, and - you know, hopefully - lead to a better software industry, because shrinkwrapped software from megafirms sucks anyway

I thought you said the sass industry and was like Jerry, no!

I'm definitely seeing the most utility with these things in "building myself bespoke and odd little software specifically for my little freak brain." Generalizing that to other people and explaining why I made those choices would be a LOT more complicated.

I'm not an LLM enjoyer but it has been realllllly good at unblocking my ADHD brain's ice wall and allowing me to start/finish my stupid little side quest projects. I also hate that this is the new norm for tech and I hate having to accept that "yelling at the computer means job security"

I totally get where you're coming from, feeling uneasy about the normalization of relying on AI for creative purposes. As a DevSecOps researcher, I'm working on projects that map and mitigate this exact exposure, and I've built a local scan https://www.tradeapollo.co/demo

So I was thinking about this a little bit, and in particular about the opensource community and ecosystem. I'm starting to think that interfaces and governance are places where there's potential leverage for people to carve out their own economic space, or at least exert leverage.

The thing is that Claude at least will reach for tools it knows, and works best when the tools it knows look like the way that the training data used them. Write something in Python with a stable library, sure! Figure out why HuggingFace's transformers and diffusers libraries are bugging, harder...

Sorry to pick on HuggingFace but they notably do tend to change internal architecture of the transformers library in ways that create breaking issues. The point is, it becomes important to have people who can negotiate stable interfaces, especially if the amount of software goes up by 100x.

It has made products for others and the barriers are not in the LLM but the human side of the equation. Strong systems thinkers can do it already and chainlink for example has many users

Crosslink, the swarm/team version is already building complex software

@dollspace.gay I was planning on testing out chainlink today. Should I go with Crosslink instead?

Have you used open local models at all?

Yes, it however lacks the hardware to try out the heavyweights in local models. It only has 24gb vram