I've been using ChatGPT-5 and it seems to have been trained on millions of PowerPoints and USA Today articles. It's bullet-happy.
It also gave me dangerous medical advice, telling me to stop a medication immediately.
Gemini is, for my purposes, far superior.
Discussion
Loading...
@jeffjarvis
I switched to Proton’s Lumo for the privacy.
I switched to Proton’s Lumo for the privacy.
Introducing Lumo: AI where every conversation is confidential | Proton
@jeffjarvis Just curious, did you think about of water and energy wasted with those tests?
Or the impact on the climate by your queries?
Or the impact on the climate by your queries?
Again, just wondering if people who use it don't know or don't care.
@jeffjarvis
Consulting humans with expertise is far superior.
@TCatInReality
Amen. Using this as a test. ChatGPT failed.
Amen. Using this as a test. ChatGPT failed.
@jeffjarvis In regards to #substack it appears we are unable to end the debate on whether or not virulent hate-spewers need to be allowed to be platformed so their hatred can be endlessly pumped into the public sphere.
A similar thing appears to be happening with the so-called “ #ai.” It’s been shown to have little or no value over and over and over again, and yet people continue to “debate” its worth.
When do these debates end?
@jeffjarvis Also, ChatGPT has always been emoji happy, but good grief. This version is Emoji's Gone Wild territory.
@jeffjarvis I’m finding Claude to be the most reliable and most likely to reconsider flawed reasoning or hallucinations. Gemini in particular seems to resist corrections and has a tendency to double down. Sounds like GPT5 may share this “arrogance.”
bonfire.cafe
A space for Bonfire maintainers and contributors to communicate
Automatic federation enabled