Thread #108299679
File: ais.jpg (88.9 KB)
88.9 KB JPG
What do you find one AI better at another AI for?
38 RepliesView Thread
>>
>>
File: 1738692550541614.webm (3.5 MB)
3.5 MB WEBM
>>108299693
>>108299679
back to shitting street please and thank you my saaaaaaaaaaaaaaaaaaars
>>
>>
chatgpt has become completely braindead due to endless optimisation for penny pinching, it has declined to just a very inefficient spell check since thats what the normies use it for
gemini has always been hit or miss, its a bit better at understanding what you want and a lot worse at everything else
>>
File: 1718675549160219.webm (1.3 MB)
1.3 MB WEBM
>>108299727
>>108299731
Oh my oh my chaat my favorite AI snack I an eating this every day with the most amazing saaaaaaaasaaars in Mumbai. Did you see recent picture of Elon Musk? He is looking so beautiful! Neurolink when? Bot boys rise up!
>>
>>
>>
>>
File: 1746518675168278.png (554.6 KB)
554.6 KB PNG
>>
>>
>>
>>
>>
>>
>>
>>108299679
Chatgpt went to shit after 5. Model routing sucks.
Google is good for images I think, and it thinks/writes too much like a pajeet.
Claude is good but you need to be aware of the tokens you use. Also for some fucking reason it doesn't allow you to change model mid conversation.
>>
>>
>>
>>
>>
>>
>>
>>
gemini is amazing for a lot of stuff, science, maths, graphics, video understanding, audio, agentic stuff.
the main drawback is that the uptime is fucking horrible and its slow.
claude is great but has shit multimodal, tokens are very expensive and it is overly verbose.
chatgpt is just very usable for daily stuff, i never run into rate limits, and codex is more like a scalpel compared to the other two, much more token efficient, but it won't explore and do as much thinking for you.
>>
>>
>>108299725
>>108299757
>>108299806
can you fuck off and fall down a well or something, tia.
>>
>>
>>
>>108299679
Gemini is good at being deranged and degenerate and some times based, ok at code. GPT is total trash, keeps wasting my time because he has to check if the code is safe and compliant before implementing. Claude is good at code, prose, but not as based as Gemini.
>>
>>108303289
yep that about sums up my experience
chatgpt seems to be more concerned about validation than anything else, as though it absolutely MUST tell me that I'm right for thinking that and I asked a great question
neither gemini nor chatgpt seem to be good at philosophical extrapolation though, since they're all gimped by censorship and guardrails anyway
>>
>>
>>
I used ChatGPT pretty much exclusively for three years, but tried Claude every now and then to compare. Recently cancelled my ChatGPT subscription after it started doing the reassuring shit all the time. Claude just feels better to talk to so I switched to that for now. I have Google One for Gemini but I just hate the way it talks and always asks follow up questions to drive engagement.
>>
Thoughts on Anthropic's free version? I'm using ChatGPT without an account for random stuff, but I'm considering switching after the whole DoW deal. The goal is to use fully local models, but I'm not quite there yet.
>>
>>
>>