3 ms·

Thankfully, there are a variety of Chinese models that never will. I think we all know that in a few years, they will also be the only relevant offerings on the

Thankfully, there are a variety of Chinese models that never will. I think we all know that in a few years, they will also be the only relevant offerings on the market, due to not being bogged down with over-zealous ""safety"" footguns.

wpietri 2mo ago
Your theory is that the Chinese government is thoroughly uninterested in safety or prosocial controls?
- dannyw 2mo ago
  Amongst Chinese labs and netizens, there's MUCH less belief/mindshare on "AGI = existential risk to humanity", "paperclip maximiser", and similar lines of thinking. AI is seen more as just a technology, and less like a scary boogyman. Whether that's right or wrong, I'll leave to you, but there's huge differences in perspectives, and if you only get your news from Western sources and communities (and companies), you're in a bubble too. A different bubble, and arguably a more porous one, but still a bubble.
  - wpietri 2mo ago
    The paperclip boogeyman is not the only reason, and probably not the biggest one, that models get safety/content constraints, however useful it is as a PR distraction. I agree Chinese models are different right now, but I think that's a function of their novelty and desire to compete globally. For a taste of where I think things are headed, try asking Chinese models about Tiananmen [1]. And then take a look at the Chinese government's approach to pretty much anything that they think reduces security or social harmony. I find it hard to believe their models will be the one exception to that over the long term. [1] https://en.wikipedia.org/wiki/1989_Tiananmen_Square_protests_and_massacre https://en.wikipedia.org/wiki/1989_Tiananmen_Square_protests...
- someguynamedq 2mo ago
  Prosocial is subjective
  - krzyk 2mo ago
    Information is prosocial.
a96 2mo ago
They already have their own telltales. Might or might not be more avoidable and varied.

wpietri 2mo ago

Your theory is that the Chinese government is thoroughly uninterested in safety or prosocial controls?

dannyw 2mo ago
Amongst Chinese labs and netizens, there's MUCH less belief/mindshare on "AGI = existential risk to humanity", "paperclip maximiser", and similar lines of thinking. AI is seen more as just a technology, and less like a scary boogyman. Whether that's right or wrong, I'll leave to you, but there's huge differences in perspectives, and if you only get your news from Western sources and communities (and companies), you're in a bubble too. A different bubble, and arguably a more porous one, but still a bubble.
- wpietri 2mo ago
  The paperclip boogeyman is not the only reason, and probably not the biggest one, that models get safety/content constraints, however useful it is as a PR distraction. I agree Chinese models are different right now, but I think that's a function of their novelty and desire to compete globally. For a taste of where I think things are headed, try asking Chinese models about Tiananmen [1]. And then take a look at the Chinese government's approach to pretty much anything that they think reduces security or social harmony. I find it hard to believe their models will be the one exception to that over the long term. [1] https://en.wikipedia.org/wiki/1989_Tiananmen_Square_protests_and_massacre https://en.wikipedia.org/wiki/1989_Tiananmen_Square_protests...
someguynamedq 2mo ago
Prosocial is subjective
- krzyk 2mo ago
  Information is prosocial.

a96 2mo ago

They already have their own telltales. Might or might not be more avoidable and varied.