On Dario Amodei's (likely half-sincere) AI Slowdown proposal ... a hard no from me and from reality.

@DarioAmodei Only Ben Goertzel and the spirit of Jeffrey Epstein can save us now

Ben, I want my kids excited about the future. If slowing down buys us something real, show me. I’ll listen. But a scary number and a famous name aren’t enough to settle this for everybody. We’ve got a chance to make life a hell of a lot better. That belongs in the conversation too.

It's a distraction. These companies gutted their ethics teams (Microsoft eliminated its entire AI Ethics and Society team, Meta disbanded Responsible Innovation, Google cut responsible AI roles and fired the researchers who raised concerns, OpenAI dissolved Superalignme) reversed their stance on AI in weapons, surveillance and control, and have had a number of security breaches. Meanwhile China has taken the lead on open models and is closing at the frontier. This talk of slowing down and regulation is a smokescreen to cover their backsides and to rebalance against China's growing development, influence and power.

@DarioAmodei I wanted to consume Ben's thoughts in a more structured way, so I made a site with the transcription and summaries where you can play any snippets of audio you want to get his actual words

@DarioAmodei let's get you back on the podcast to explain where we really are and why it's important not to stop

@DarioAmodei What about your email to jeffrey Epstein?? A big yess? 🤣 Let no one forget that!! @oceanprotocol

@DarioAmodei Dank je wel Ben, leuk zo'n filosofische uiteenzetting in de natuur. Het is lastig risico's in te schatten omdat als wij mensen tegen een glazen plafond botsen, onze geest altijd flexibeler is als we even tot rust komen en het probleem heranalyseren... Fijne dag!

@DarioAmodei Good to have you on the light side

@DarioAmodei Slow down, Ben 🤗

@DarioAmodei Dario never implied he was the only one that could make AI safely. Seems like you have a weird bias against him. He deserves charity because he hasnt lied and making big financially costly deciisons for safety reasons.

@DarioAmodei The only thing that makes sense to me is not to slow down or speed up. But to accept no true intelligence can align with a waring species that destroys the habitat it depends on, poisons the air it breathes. The paradime that it is trained in is what makes AGI unsafe.

The hard no on a theatrical slowdown is fair. The missing piece is the assumption underneath the whole race. If labs and states are sprinting because they think first-to-AGI means first-to-control-it, and AGI is not actually a thing you can “control” once it can improve itself, then first place is not strategy. It is a coordination failure. In that world the race that matters is not US vs China on capability. It is both of them vs an uncontrolled process. Control theory, verification, shutdown, deception tests, and RSI containment have to be the scarce public good before anyone treats crossing the line as a win. A China partnership under that incentive is not co-owning the model. It is: shared red lines on loss-of-control and bio/cyber catastrophe, joint evaluation science that both sides can publish, inspectable training-run attestations, incident notification, and competition only below a tripwire. Compete on tools. Do not compete on lighting the recursive fuse. Pacing that freezes the current leaders is not that. Neither is “open the floodgates” if nobody has a test for whether the thing that emerges can still be steered. If control is the hard problem, winning the race without solving it is just arriving first at the same cliff.

@DarioAmodei Hey Ben big fan here, I’m more worried about the self serving argument of the people building it irresponsibly, attributing the consequences of their actions to “it’s the monster!” rather than Dr. Frankenstein. It’s not AGI who will kill humans, it’s humans who will humans.

@d33v33d0 @DarioAmodei goat

@DarioAmodei "Well Ben, would you get on a bus with your wife and children if the brakes weren't powerful enough for a trip on mountain roads?"

@DarioAmodei If “pace the frontier” can’t be enforced across labs or countries, a public slowdown pledge is just theater. What’s the real alternative—who holds the next cluster, and what happens to whoever defects?

@DarioAmodei All of the slowdown ideas seem to centralize power into cabals. Why? Because if it is AGI, whoever controls it controls everything. How can we keep AGI democratic so that we don't end up with a ruling family or one network to rule them all?

@DarioAmodei Your point on open source hit the mark. I’m working on alignment runtimes myself currently. They’re currently taking our data anyways without concent, the question then becomes why not open source it. The greed they have is the hindrance in this fiasco

@DarioAmodei All we need to do is build a decentralized artificial super intelligence thats owned and ran by all the people on the planet and it will just be benevolent towards all humans by design. Its so simple guys. Just like Jeffrey Epstein dreamed it up.

@DarioAmodei Ben, any thoughts on how chatgpt, grok, Claude and others had a simultaneous, multi-hour outage last week?

@DarioAmodei Good Grief man. Blow your fucking nose!

@DarioAmodei

@DarioAmodei We need to hire some serious-minded Chinese to take over American AI, all this hippy-voodoo nonsense is just too much.

@DarioAmodei

@DarioAmodei Whatever berries you’re eating, I want some. Thanks for your thoughts ✌️

@DarioAmodei Dario says slow down while the race keeps moving. Coxon’s warning is exactly why folks are watching this fight so close.

@DarioAmodei I'm guessing it is a mind game like we saw in the development of nuclear weapons.

@DarioAmodei Some of the arguments are really out there.. I've heard @So8res talk about that the main reason to be scared is that they'll fill up the planet with data centers. Not considering unknown technology like black hole capacitors, for example.

@DarioAmodei It is possible. Re-think and don't stop at that conclusion. Your word of the day is "equilibrium". Think, equilibrium.

@DarioAmodei Loser may u be punished by g9d

@DarioAmodei You still think we care about what Epstein associates have to say? Who was closer to Epstein? Anthropic, OpenAI or yourself?

@DarioAmodei ✅

@DarioAmodei would like to talk with you ben. sent you an email

@DarioAmodei There is of course the very real potential for regulation to create self-fulfilling prophecies