Skip to main content

Anti-AI technologists only:
If and when the AI bubble finally collapses due to financial strain, and LLMs become synonymous with several open-source, open-weight, ethically trained models that run on your modest local hardware, would you use it?

This is a world without OpenAI, Anthropic, and those massive AI data centres. Maybe some smaller shops spin up no-code wrapper tools, but nothing like the surveillance capital shit we have today.

Basically, if the ethics, ecology, and extractive capitalism bits go away, would you consider working with these tools?

#poll

  • No. (42%, 48 votes)
  • I'd consider trying it. (28%, 32 votes)
  • It'd be another open-source library to me. (8%, 10 votes)
  • I'd actively build with it. (8%, 9 votes)
  • Other, comment. (11%, 13 votes)
112 voters. Poll end: 2 weeks ago

#poll
Hypolite Petovan
Hypolite Petovan  
•
@May Likes Toronto No, I don't use natural language to talk to machines. I've learned several languages that are meant to convey deterministic commands to a computer and that's a hill my career of software engineer might end up dying on.
@May Likes Toronto
Hypolite Petovan
Hypolite Petovan  
•

@silverwizard @May Likes Toronto I've seen meta discussions around using Claude that involves creating a structured grammar to condense the context to minimize the token expense for a given query.

We're reinventing code to interface with a nondeterministic commercial entity and I'm not happy about it. Even when you strip the "commercial" aspect, you're still left with a nondeterministic system and while it works wonders for bounded problems, it's terrible when applied to open ended questions and not being able to balk when internal output confidence level is too low, none of which depend on the current unethical externalities.

@silverwizard @May Likes Toronto
⇧