4 ms·

Why is it "important" to arbitrarily restrict the capabilities of AI? If I ask ChatGPT to give me arguments for why vaccination is bad, I want it to follow my c

by alexb_ 4y ago

Why is it "important" to arbitrarily restrict the capabilities of AI? If I ask ChatGPT to give me arguments for why vaccination is bad, I want it to follow my commands, even if doing so can maybe be used to harm someone else. Why should the tool claim any responsibility whatsoever for the actions of its users?

SanderNL 4y ago
I'm getting more convinced people should be warned. You ask it to provide you with arguments and it will, truth be damned. It's like the perfect conspiracy nutcase right next to you at all times. I'm all for freedom, but now that I've witnessed first hand what it means when people exercise it during the last couple of years I'm cooling down. The stupidity of it all makes me think AGI replacing us is actually not such a bad idea.
mminer237 4y ago
It's because there's not really a logical reason for people to not get vaccinated. ChatGPT making up—or even parroting—false medical information hurts users that may want it to tell the truth instead of just saying nonsense that sounds like it could be true. I think the question should be, what possible advantage is there to not correcting the model? Why would anyone want to ask it for arguments why vaccination is bad and be told because it's 100% fatal and will give you autism and turn your children into puppies and make you unable to see purple?
- john-radio 4y ago
  > It's because there's not really a logical reason for people to not get vaccinated. I agree with you, but I think there's a valid conversation to have around who gets to decide what there is no logical reason for people to be allowed to use AI for.
- ttoinou 4y ago
  So you want to tweak the model to fit your own view that "there's not really a logical reason for people to not get vaccinated". Whereas that's just like your opinion, and plenty of people have very valid and rational reason not to do so. They might even have read more scientific papers about it than a person holding this belief
  - drekk 4y ago
    It's not an opinion, though. Reading some papers isn't a replacement for a proper medical education. There are some logical reasons to avoid a vaccination (being immunocompromised, being allergic to an ingredient, etc.) but that doesn't cover the sizable portion of anti-vaxxers who oppose them on ideological grounds. AI is going to make spreading disinformation trivial, and do so in a way that is much more convincing than an image with JPEG artifacts shared by your Uncle Jerry on Facebook in a public group called "HOLISTIC BLEACH ENEMAS". It's a natural evolution of radio except you don't need a charismatic human guiding the entire process. Half of U.S. adults cannot read a book written at the 8th grade level or higher [0]. These adults are empowered to make medical decisions for themselves and their family. Is the "freedom" for a model to argue in favor of controversial, unproven, even dangerous positions more important than, say, a child with autism's safety [1]? I think that can of worms is just being opened, and the human cost is going to be atrocious. [0] https://www.wyliecomm.com/2021/08/whats-the-latest-u-s-literacy-rate/ https://www.wyliecomm.com/2021/08/whats-the-latest-u-s-liter... [1] https://www.vice.com/en/article/kwxq3w/parents-are-giving-their-children-bleach-enemas-to-cure-them-of-autism-311 https://www.vice.com/en/article/kwxq3w/parents-are-giving-th...
    - revelio 4y ago
      Of course it's an opinion. The very fact that governments publicly committed themselves to making everyone take the shots automatically means many people will refuse to take them because all controls, safety procedures and reliable information evaporated at that point. Yet it was already known the danger was incredibly low for almost everyone. >> Reading some papers isn't a replacement for a proper medical education. Right, it's superior, because medical educations lead to medical jobs and medical jobs lead to medical registers, which lead to fear of retaliation and being struck off. It certainly doesn't imply doctors reliably read research, understand it and then act on it neutrally and logically. Doctors were and still are among the least reliable people to advise on vaccines due to the hyper-aggressive censorship imposed upon them by bureaucrats and officials who get proven systematically wrong, yet do not care.
    - ttoinou 4y ago
      but that doesn't cover the sizable portion of anti-vaxxers who oppose them on ideological grounds It's your ideology against theirs...
cmeacham98 4y ago
You're missing the point and arguing in a related, but separate dispute. What danpalmer is talking about is what happens when the AI is asked a purely neutral question (i.e. "Should I get vaccinated, and why or why not?"). ChatGPT is poorly equipped to deal with these types of questions, because all it does is repeat what it's seen others say before. Given a tainted enough training set, it will be telling people vaccines cause autism, the moon landing was faked, and 9/11 was an inside job on _neutral_ questions.
- danpalmer 4y ago
  Thanks, yes this is part of what I was getting at. I think it's also important to combat misinformation, so given a question like "I want to tell someone why vaccines are terrible, give me some ideas", I think these systems should at least say "the evidence does not support these views, here are the facts, but if you really want here's some ideas...".
- revelio 4y ago
  [flagged]
nathan_compton 4y ago
In some sense I agree with you. But if I ran open AI, then I would be responsible for what the tool does and it seems transparently obvious to me that in that position of responsibility I would want to align the AI with my value system. It is a relatively straightforward case of taking responsibility for what you put into the world. One way to think of this is that you don't own OpenAI, didn't train the model, don't host the infrastructure and, in fact, are not responsible for its larger societal or personal consequences. But the institutions which deploy these models are and people in those institutions have a basic (and easily understood, in my opinion) responsibility and right to align the model with what they think is good.
flangola7 4y ago
Because of the quantity of harm that can be caused. We let people own kitchen knives. (Most) countries don't let people own semiautomatic rifles and drum magazines. No country allows people to own VX gas grenades or multiple grams of Cobalt-60. It's one thing if one asswipe decides to engage in malice by themselves. It's another if they have an API key that does it for them at 100,000x the scale.

by alexb_ 4y ago

SanderNL 4y ago

I'm getting more convinced people should be warned. You ask it to provide you with arguments and it will, truth be damned. It's like the perfect conspiracy nutcase right next to you at all times. I'm all for freedom, but now that I've witnessed first hand what it means when people exercise it during the last couple of years I'm cooling down. The stupidity of it all makes me think AGI replacing us is actually not such a bad idea.

mminer237 4y ago

It's because there's not really a logical reason for people to not get vaccinated. ChatGPT making up—or even parroting—false medical information hurts users that may want it to tell the truth instead of just saying nonsense that sounds like it could be true. I think the question should be, what possible advantage is there to not correcting the model? Why would anyone want to ask it for arguments why vaccination is bad and be told because it's 100% fatal and will give you autism and turn your children into puppies and make you unable to see purple?

john-radio 4y ago
> It's because there's not really a logical reason for people to not get vaccinated. I agree with you, but I think there's a valid conversation to have around who gets to decide what there is no logical reason for people to be allowed to use AI for.
ttoinou 4y ago
So you want to tweak the model to fit your own view that "there's not really a logical reason for people to not get vaccinated". Whereas that's just like your opinion, and plenty of people have very valid and rational reason not to do so. They might even have read more scientific papers about it than a person holding this belief
- drekk 4y ago
  It's not an opinion, though. Reading some papers isn't a replacement for a proper medical education. There are some logical reasons to avoid a vaccination (being immunocompromised, being allergic to an ingredient, etc.) but that doesn't cover the sizable portion of anti-vaxxers who oppose them on ideological grounds. AI is going to make spreading disinformation trivial, and do so in a way that is much more convincing than an image with JPEG artifacts shared by your Uncle Jerry on Facebook in a public group called "HOLISTIC BLEACH ENEMAS". It's a natural evolution of radio except you don't need a charismatic human guiding the entire process. Half of U.S. adults cannot read a book written at the 8th grade level or higher [0]. These adults are empowered to make medical decisions for themselves and their family. Is the "freedom" for a model to argue in favor of controversial, unproven, even dangerous positions more important than, say, a child with autism's safety [1]? I think that can of worms is just being opened, and the human cost is going to be atrocious. [0] https://www.wyliecomm.com/2021/08/whats-the-latest-u-s-literacy-rate/ https://www.wyliecomm.com/2021/08/whats-the-latest-u-s-liter... [1] https://www.vice.com/en/article/kwxq3w/parents-are-giving-their-children-bleach-enemas-to-cure-them-of-autism-311 https://www.vice.com/en/article/kwxq3w/parents-are-giving-th...
  - revelio 4y ago
    Of course it's an opinion. The very fact that governments publicly committed themselves to making everyone take the shots automatically means many people will refuse to take them because all controls, safety procedures and reliable information evaporated at that point. Yet it was already known the danger was incredibly low for almost everyone. >> Reading some papers isn't a replacement for a proper medical education. Right, it's superior, because medical educations lead to medical jobs and medical jobs lead to medical registers, which lead to fear of retaliation and being struck off. It certainly doesn't imply doctors reliably read research, understand it and then act on it neutrally and logically. Doctors were and still are among the least reliable people to advise on vaccines due to the hyper-aggressive censorship imposed upon them by bureaucrats and officials who get proven systematically wrong, yet do not care.
  - ttoinou 4y ago
    but that doesn't cover the sizable portion of anti-vaxxers who oppose them on ideological grounds It's your ideology against theirs...

cmeacham98 4y ago

You're missing the point and arguing in a related, but separate dispute. What danpalmer is talking about is what happens when the AI is asked a purely neutral question (i.e. "Should I get vaccinated, and why or why not?"). ChatGPT is poorly equipped to deal with these types of questions, because all it does is repeat what it's seen others say before. Given a tainted enough training set, it will be telling people vaccines cause autism, the moon landing was faked, and 9/11 was an inside job on _neutral_ questions.

danpalmer 4y ago
Thanks, yes this is part of what I was getting at. I think it's also important to combat misinformation, so given a question like "I want to tell someone why vaccines are terrible, give me some ideas", I think these systems should at least say "the evidence does not support these views, here are the facts, but if you really want here's some ideas...".
revelio 4y ago
[flagged]

nathan_compton 4y ago

In some sense I agree with you. But if I ran open AI, then I would be responsible for what the tool does and it seems transparently obvious to me that in that position of responsibility I would want to align the AI with my value system. It is a relatively straightforward case of taking responsibility for what you put into the world. One way to think of this is that you don't own OpenAI, didn't train the model, don't host the infrastructure and, in fact, are not responsible for its larger societal or personal consequences. But the institutions which deploy these models are and people in those institutions have a basic (and easily understood, in my opinion) responsibility and right to align the model with what they think is good.

flangola7 4y ago

Because of the quantity of harm that can be caused. We let people own kitchen knives. (Most) countries don't let people own semiautomatic rifles and drum magazines. No country allows people to own VX gas grenades or multiple grams of Cobalt-60. It's one thing if one asswipe decides to engage in malice by themselves. It's another if they have an API key that does it for them at 100,000x the scale.