6 ms·
The main thing that ChatGPT has gotten better at is rejecting jailbreaks and refusing to go off the reservation. It has been demonstrated that "safety" trades o
by AbrahamParangi 3y ago
The main thing that ChatGPT has gotten better at is rejecting jailbreaks and refusing to go off the reservation. It has been demonstrated that "safety" trades off against "capability". I'm sure OpenAI has evaluations that demonstrate the improvements they've been making to "safety" have not come at the cost of capability, but I'd bet those evaluations are wrong (by being insufficient). It also wouldn't surprise me if the tradeoff between "safety" and capability is just intrinsic.
You can have your model talk like HR or think like a mad scientist, but not both equally well.
- spaceman_2020 3y agoI've cancelled my pro subscription. You can't ask me to pay money and also expect me to spend hours Googling around to find prompts for basic queries. I would also like to not be treated like an idiot. Every time I query anything related to health or medicine, chatGPT will give me a short generic answer, then add two paragraphs of warnings about how I should just seek a medical professional's help. As if I'm the kind of moron who will blindly follow whatever some AI tool tells me and not actually go to a doctor if there's something wrong with me. This organization seems like it's being run by (scared) lawyers.
- copperx 3y agoI'm nostalgic about the first public version of GPT-4. It did have some sparks of AGI. The interface was very ergonomic because it assumed correctly many things about the user's request and intentions. Now I find I have to write very long prompts and explanations to steer it correctly, as if training a brand new employee. It's really tiring. And I'm not even talking about the accuracy of the responses. It's sad, but I understand why they did it. It was too much power in the hands of people.