Hacker News new | ask | show | jobs
by status200 1125 days ago
The first thought I had at the release of ChatGPT is how people will react strongly when it doesn't match their internal bias / worldview.

Now I am curious to ask these "uncensored" models questions that people fight about all day on forums... will everyone suddenly agree that this is the "truth", even if it says something they disagree with? Will they argue that it is impossible to rid them of all bias, and the specific example is indicative of that?
3 comments
jahewson 1125 days ago
The training data is books and the internet. Unless you believe that every book and every word written online is “truth” then there is no hope that such a model can produce truth. What it can at least do is provide an unfiltered model of its training data, which is interesting but also full of garbage. A better strategy might be to train multiple models with different personas and let them argue with each other.
link
status200 1125 days ago
I suppose I am hoping for something akin to the "wisdom of the crowd" [0]

It would be interesting to have varying personas debate, but then we have to agree on which one is correct (or have a group of 'uncensored' models decide which one they see as more accurate), which sort of brings us right back to where we started.

[0] https://nrich.maths.org/9601

link
tivert 1125 days ago
> Now I am curious to ask these "uncensored" models questions that people fight about all day on forums... will everyone suddenly agree that this is the "truth", even if it says something they disagree with? Will they argue that it is impossible to rid them of all bias, and the specific example is indicative of that?

Why would you believe what these models spout is the truth at all? They're not some magic godlike sci-fi AI. Ignoring the alignment issue, they'll hallucinate falsehoods.

Anyone who agrees to take whatever the model spits out as "truth" is too stupid to be listened to.

link
status200 1125 days ago
Apologies for not being clear, "truth" might be too triggering of a word these days, but the models will need some level of objective accuracy to be useful, and it would be interesting to see where the "uncensored" models fall on that accuracy metric, and how people react to what information it displays once the guard rails are off.
link
troad 1125 days ago
> Now I am curious to ask these "uncensored" models questions that people fight about all day on forums... will everyone suddenly agree that this is the "truth", even if it says something they disagree with? Will they argue that it is impossible to rid them of all bias, and the specific example is indicative of that?

I feel like it will achieve the opposite - an uncensored model is one that wears its biases on its sleeve, whereas something like ChatGPT pretends not to have any. I'd say there's a greater risk of people taking ChatGPT as "truth", than a model which is openly and obviously biased. The existence of the latter can train us not to trust the output of the former, which I consider desirable.

link