it’s “self-interested” in that i picked my side intentionally because i believe it is the right side to be on. i’ve had plenty of opportunities to join the other side. i’m picking the side with less capital because i want to live in a world where this stuff is open.

the models in the competitive world are less powerful and more controllable. you aren’t going to convince me to trust dario that he’s gonna make sure everything is okay.

it’s a tweet. “the work i do to facilitate open model research” implies i believe that open model research is a good thing. i do. you know this. we’ve talked about it before. you’re being pedantic.

the open model world is a race to the bottom on price. this is good for limiting the pace of capability gain.

the most antisocial and competitive company in this race is anthropic. they are frequently alone at the frontier. they do not collaborate. they cut off anyone they see as a threat to their dominance or pricing power.

i trust jensen, then demis, then elon, then sam, then dario.

@tautologer it is the first publicly available model that i am explicitly not allowed to use for my work, because anthropic holds the view that the work i do to facilitate open model research is harmful. capability and alignment research are coupled. anthropic wants to be the only lab.

@willccbb @tautologer how dare you help people build better product sir

@tautologer @willccbb @badlogicgames “Fable 5 will not fall back to a different model. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT).”

I mean this is a pure self interest argument independent of whether the work is actually harmful or not

A multipolar or competitive environment with extraordinarily powerful models would be horrifying. We see this on a smaller scale with image gen & misinformation already.

Anarchy or nihilism is far more dystopian than anything even the worst authoritarians could come up with. It matters less who is the only lab and more that there is an only lab

@willccbb @badlogicgames @tautologer Apparently they’ll let you use it and silently sabotage your systems

@potencytoact @willccbb @badlogicgames no they won't, they clearly state it will attempt to help the user, just with degraded capability. big difference!

@_ueaj @tautologer nobody’s launched a nuke since the unipolarity days.

central authorities best serve the interests of the people when their authority isn’t guaranteed.

@tautologer @willccbb @badlogicgames your confusion is proof that you are confused

@willccbb @tautologer “Nobody can research AI, it’s too dangerous!! Except for me, I’ll handle it responsibly.”

The specific argument in the comment you made is purely self interested though. There is no mention or argument against monopolarity in it.

Regardless, no, the models in the competitive world are not more controllable, they are subject to forces and incentives that encourage a race to the bottom. Even right now companies are not really "in control" of themselves, motivated reasoning and selective pressures move them in ways they can't comprehend.

A world without trust is very appealing to a lot of people, but there really is no substitute. Compromise is a better path forward than antisocial competition

nuclear weapons were destined to exist in any sufficiently long-lived society, just like large language models. these are discoveries not inventions.

technology will always progress, and benefit those in power when centralized. broad open diffusion of technology is a check on power. the internet and smartphones and VPNs and open models are powerful levers against authoritarian leaders.

@potencytoact @willccbb @badlogicgames yes that is what i said and not what you said

@potencytoact @willccbb @badlogicgames sabotaging _the quality of the reply they give you_ is not the same as sabotaging _your systems_

giving a _less useful reply than they otherwise could_ is not the same as _giving an actively harmful reply_

We have only had nuclear weapons for ~80 years. Only one nuclear power has collapsed and it was a miricle it didn't end in catastrophe. I don't think we should gamble humanity's fate on not doing this for the next 1000 years or whatever.

Whatever system you can envision can be done better by a machine, more accountable, more etc., unless you don't believe in material consciousness or superintelligence in general which I don't think is true.

Regardless, it has become harder, not easier, to overthrow or hold the government accountable as technology has progressed.

@willccbb @samsja19 @tautologer Anthropic needs to be shattered now. Dario needs to be put in his place. And FORCED to hold hands with @sama