My hot take: gpt-5 might be a worse product AND that might be a good thing. Anthropomorphizing a technology with an already sort-of-anthropomorphic UI can lead people in bad directions (e.g. see the wave of AI psychosis), so I’m happy OpenAI is stepping off the ledge a bit with that one
Using LLM *should* feel like using a computer rather than talking to a person.
It definitely feels better for more technical related questions or analyses, but it should at least allow the user to maybe adjust for personality when it comes to more personal or non-technical discussions.
My hot take: gpt-5 might be a worse product AND that might be a good thing. Anthropomorphizing a technology with an already sort-of-anthropomorphic UI can lead people in bad directions (e.g. see the wave of AI psychosis), so I’m happy OpenAI is stepping off the ledge a bit with that one
Using LLM *should* feel like using a computer rather than talking to a person.
It definitely feels better for more technical related questions or analyses, but it should at least allow the user to maybe adjust for personality when it comes to more personal or non-technical discussions.
A customizable experience may be ideal for most