121–130 of 323 posts
Earlier quoted context omitted.
How can you be so sure? I've seen a documentary that detailed the experiences of a prince from abroad working in fast food after being sent to the US to get some life experience before getting married. Maybe it's more common than you think.
Prince Akeem of the nation of Zamunda! :D
Okay this is just getting suspicious. Their excuses for keeping the chain of thought hidden are dubious at best [1], and honestly just seemed anti-competitive if anything. Worst is their argument that they want to monitor it for attempts to escape the prompt, but you can't. However the weirdest is that they note that: > for this to work the model must have freedom to express its thoughts in unaltered form, so we cann…
Occam's razor: there is no secret sauce and they're afraid someone trains a model on the output like what happened soon after the release of GPT-4. They basically said as much in the official announcement, you hardly even have to read between the lines.
The best part is that you still get charged per token for those CoT tokens that you're not allowed to ask it about.
Would be funny if there was a human in the loop that they're trying to hide
Would be funny if there was a human in the loop that they're trying to hide
In the early days of Google, when I worked on websearch, if people asked me what I did there, I'd say: "I answer all the queries that start with S."
I remember bumping into a very famous US politician in the lobby and pointing that marquee out to him just as it displayed a particularly dank query.
Okay this is just getting suspicious. Their excuses for keeping the chain of thought hidden are dubious at best [1], and honestly just seemed anti-competitive if anything. Worst is their argument that they want to monitor it for attempts to escape the prompt, but you can't. However the weirdest is that they note that: > for this to work the model must have freedom to express its thoughts in unaltered form, so we cann…
Or, without the safety prompts, it outputs stuff that would be a PR nightmare. Like, if someone asked it to explain differing violent crime rates in America based on race and one of the pathways the CoT takes is that black people are more murderous than white people. Even if the specific reasoning is abandoned later, it would still be ugly.
Okay this is just getting suspicious. Their excuses for keeping the chain of thought hidden are dubious at best [1], and honestly just seemed anti-competitive if anything. Worst is their argument that they want to monitor it for attempts to escape the prompt, but you can't. However the weirdest is that they note that: > for this to work the model must have freedom to express its thoughts in unaltered form, so we cann…
We know for a fact that ChatGPT has been trained to avoid output OpenAI doesn't want it to emit, and that this unfortunately introduces some inaccuracy.
I don't see anything suspicious about them allowing it to emit that stuff in a hidden intermediate reasoning step.
Yeah, it's true they don't what you to see what it's "thinking"! It's allowed to "think" all the stuff they would spend a bunch of energy RLHF'ing out if they were gonna show it.