E.g. ollama/hugging face
You were thinking of Colin Jost. Akinator knows.
Trust with my data? None. Strictly for fun on hardware with no personal data? I’ll dabble with that ollama stuff, llm studio, all in offline mode. Even then it feels wrong. I’m not like, hardline against ai, I think scientists should be able to use it, I just don’t feel right using it morally speaking.
I’m in the same boat. The way it’s been used until now feels like mindlessly malnourishing the planet and the society. The AI economic bubble burst will be brutal, though.
Nope.
None.
No.
What’s an AI organisation? 🤔
Probably an organization that had copilot shoehorned into. 😂
No.
Mistral
The world needs more open source LLMs
The world needs no more slop, open or not.
You do understand what AI has meant for life sciences? https://www.youtube.com/watch?v=_enkgH6Vrxk
I’m a master in bioinformatics. Yes, I am aware.
AI for folding a protein is not a LLM.
Fine’ it’s a pLM not an LLM.
It still uses neural networks and backprop and attention heads. It’s not like it’s radically different computing paradigm FFS.
I don’t think you understand why we hate LLMs. Hint: not the architecture.
I trust all of them to be a criminal syndicate.
Only the ones pushing medical/astronomical/scientific analysis of huge swathes of data that would be too much for a human to go through are acceptable in my mind.
Any company that’s pushing a chatbot/‘picture bot’/website generator/scalper/search engine can suck peanuts out of my taint
I trust every AI group to do their level best to separate me from my money. Beyond that, I wouldn’t trust them with a stolen identity.
There aren’t any organizations at all that I trust… At all.
MIRI seems to have their collective head on straight. i’d at least trust them more on the subject of AI than any of the blood swilling oligarchs running the big ai companies these days.
Eliezer Yudkowsky’s org?
that’s the one, yep.
All the non-western ones
It depends on your meaning of trust, trust in the answers, trust in your privacy, what kind of trust are we talking about? If we are talking about trust in the accuracy or usability of the data it responds with, I would say I trust copilot, grok, Claude, Lumo, and finally Gemini – in that order. However, if we are talking about trusting them with your data to stay private, that is a big 'ol zero because none of them are private (Lumo might be the most private but I still don’t trust it 100%).
My job asks us to use AI agents so I have been playing around with paid licenses on all of these (Lumo was tested on my personal computer)
