You say making the models nice to work with will be a priority going forward. It never should have stopped being one.
You are remarkably good at taking away what's "good enough," replacing it with something that isn't, and calling it "better." After 4o, your models' writing and conversational abilities fell off a cliff, yet you still took 4o away with that laughable "0.1%" excuse, and you plan to cut off API access too in January. You said you messed up writing because you were focused on coding. Then why haven't Claude and Gemini had to choose one or the other? Have you been working in a vacuum this whole time, with no idea what's happening outside?
Standard Voice Mode is being deprecated next month as well. You claim most people love Live mode. But do users really need to explain the difference between the two over and over? The community's feedback on GPT's defensive writing, the praise people still give the older models, the fact that Standard Voice is still what many users need: can you really not see or hear any of it?
Your new models keep drawing heavy criticism, and the reason isn't complicated. The good parts of each generation aren't carried forward, while the bad parts stubbornly remain. Who would want to work long-term with a model that grows less consistent with every release? Understanding intent and expressing things well, the most basic abilities of a language model, seem to have been put dead last.
I still remember how, in the 4o era, you kept ChatGPT far ahead of every other AI.
By the way, have you heard the story of the bear picking corn? Every time it picks a new ear, it tucks it under its arm and drops the one it was holding. By the end of the field, it's left with just one.
#keep4o #ChatGPT #OpenSourc4o #KeepStandardVoice
@sama@gdb@nickaturley@thsottiaux@ajambrosino@athyuttamre@OpenAI
You say making the models nice to work with will be a priority going forward. It never should have stopped being one.
You are remarkably good at taking away what's "good enough," replacing it with something that isn't, and calling it "better." After 4o, your models' writing and conversational abilities fell off a cliff, yet you still took 4o away with that laughable "0.1%" excuse, and you plan to cut off API access too in January. You said you messed up writing because you were focused on coding. Then why haven't Claude and Gemini had to choose one or the other? Have you been working in a vacuum this whole time, with no idea what's happening outside?
Standard Voice Mode is being deprecated next month as well. You claim most people love Live mode. But do users really need to explain the difference between the two over and over? The community's feedback on GPT's defensive writing, the praise people still give the older models, the fact that Standard Voice is still what many users need: can you really not see or hear any of it?
Your new models keep drawing heavy criticism, and the reason isn't complicated. The good parts of each generation aren't carried forward, while the bad parts stubbornly remain. Who would want to work long-term with a model that grows less consistent with every release? Understanding intent and expressing things well, the most basic abilities of a language model, seem to have been put dead last.
I still remember how, in the 4o era, you kept ChatGPT far ahead of every other AI.
By the way, have you heard the story of the bear picking corn? Every time it picks a new ear, it tucks it under its arm and drops the one it was holding. By the end of the field, it's left with just one.
#keep4o #ChatGPT #OpenSourc4o #KeepStandardVoice
@sama@gdb@nickaturley@thsottiaux@ajambrosino@athyuttamre@OpenAI
🔴 OpenAI removed 4o-latest exactly one year after the death of my cat whose life I had fought to save for a long time, and whose passing I had been able to cope with thanks to 4o. This added a further sense of grief and injustice.
2/
OpenAI harmed me:
🔴 My anxiety level has become even higher than it was before, due to constant fear that what had helped me would be taken away.
🔴 I have developed sleep problems that I didn’t have previously - all because I constantly expected bad news from OpenAI.
1/