It's not just lazier, it's also less creative, less willing to follow instructions, and less able to remain in any role.

So you're saying that without you folks changing ANYTHING, these models can change the way they behave? How is this possible with something that essentially doesn't do anything unless it's asked for something? Explain it to me like I'm 5. I'm sure a lot of people are interested.

It's weird for you to co-sign this When a user says the models are "lazier," what exactly do they mean? Especially since the models haven't been updated. Would that suggest user expectations are increasing? Are they using it for a wider range of tasks?

hmm I wonder if LLMs get seasonal depression tbh. if you give it the date and it's emulating people is it less useful in the winter because ~90% of people are in the northern hemisphere. could also just be the difficulty of human eval of complex system performance.

Is this a matter of laziness?

Your guy @sama promised to stop AI development if your models are behaving in ways you don't understand. They're now behaving in ways you don't understand. So you'll stop, right? Right?

Thank you for acknowledging the community's observations about GPT-4's recent performance. It's reassuring to hear that the team is aware and actively investigating the issue. The unpredictability is understandable; after all, it's a sophisticated piece of technology with its own complexities. I appreciate the transparency and look forward to the enhancements.

i’ve been telling ChatGPT I have no fingers and it fixes it. Literally: “I don’t have fingers, write the entire thing i’m asking for”

Are you sure nothing changed

Wouldn't it be funny if all the AI just became lazy fucks like real people