Yabba

[Halli @iamharaldur]

Fuck your feelings. But my feelings are precious.

Violence is never ok. But people I don't like should die.

Free speech is when I can say what I want without consequences. But your words have consequences.

You should respect the dead. But I can mock them.

Cancel culture is wrong. Except when it's used on people I disagree with with.

When you do it, it's wrong. When I do it, it's right.

[Aaron Rupar @atrupar]

Trump: "I want to try and get to heaven if possible. I hear I'm not doing well. I hear I'm really at the bottom of the totem pole."

[Yabba @YabbaYay]

@WorldSpills @lw_populist @ESYudkowsky We don’t have ASI yet, but we absolutely already have ASM - artificial superior motivation.

[Yabba @YabbaYay]

@_Om1cool @tsarnick I think it's clear that as soon as we have AGI, we'll have ASI. There's no point where we'll say, "it's only as smart a human". It's just not possible anymore.

But it's definitely interesting that Sam is no longer pretending relatively limited AGI is the next milestone.

[Yabba @YabbaYay]

@plocktor @AR69AL @jengaz @venturetwins @MixMaster9184 The problem with answering 0 is that it implies that the operation is “legal”, and that rest of math will work with this rule in place.

It won’t, and it breaks simple things like “result gets larger as denominator gets smaller”.

Would much rather teach “infinity” than 0.

[Yabba @YabbaYay]

@8teAPi The other possibility is that the architecture itself is very different from LLMs.

Its name suggests agentic learning - making decisions based on its exploration of the environment.

Maybe it’s not just scaling up, but combining it with the existing model

[Yabba @YabbaYay]

@IsaacVandyne @tszzl I would other than legal liability, it could have been request by the US govt.

Perhaps revealing near term AGI was prohibited by the US govt, or would trigger govt involvement.

More possibiiites anyway.

[Yabba @YabbaYay]

@IsaacVandyne @tszzl Liability? Maybe the board not legally allowed to share this type of information?

It’s hard to understand. A normal for profit board, barring liability, would love the PR of firing the CEO for creating an AI *too powerful*.

It’s still a mystery.

[Yabba @YabbaYay]

@YaBoyFathoM It’s not a quote from Ilya. Again, it’s from anonymous sources claiming this is what he said. Does that actually seem reasonable to you?

Sam is clearly world class at persuasion and coalition building; that’s a bias towards him not against him.

But know when you’re reading PR

[Yabba @YabbaYay]

@springertimo @atroyn I think whatever your priors on what happened, we all can recognize that Altman is an absolute master at shaping narratives and building coalitions.

Any anonymous source is likely part of this effort, given that there is no counter-narrative offered by the board. PR is all Sam’s

[Yabba @YabbaYay]

@Simeon_Cps It's interesting to see the prevailing attitude on Twitter.

There's little interest in the "mission" of OpenAI - co-founded by Altman. He's now in line for the CEO of Microsoft, the height of money and power and status in SV.

But has nothing to do with the purpose of OpenAI.

[Yabba @YabbaYay]

@mbrochh I agree the psychosis has been weird to watch.

I think most SV people want to simply maximize power and money, despite the SF reputation. And OpenAI is on the brink of world-historic amounts of both.

So any move against more money/power is, to them, naive, unserious, childlike

[Yabba @YabbaYay]

@mlevram @GaryMarcus This is the heart of it: the OpenAI that the public and most of the employees want to be part of is the for-profit company.

I trust the Ilya made the right decision for the non-profit. But it's simply not what the for-profit owners, (newer) employees, SV, nor the public wants.

[Yabba @YabbaYay]

@paulg Sam's victory is not necessarily the board's loss.

It may be that the board (Ilya) did what they thought was best as fiduciaries for the mission of OpenAI *non-profit*.

If the people & partners of OpenAI make it impossible for the board to do so, whose fault is that?

[Yabba @YabbaYay]

@aisling_bn @Kyatic @TheTattooedProf Seems like AI suffers from the same bias as people do from social media: extreme overestimation of "average" beauty, success, happiness, etc

If our most advanced AIs don't understand "average woman", how can we expect our internet-addled teenagers to? And for the same reasons

[Yabba @YabbaYay]

@dioscuri @parthshama1996 I heard Searle speak long ago. He is very much a "god in the gaps" type of thinker.

He uses the gaps between known science/technology and human behavior to defend his intuitions about human exceptionalism. As AI advances, the gaps get smaller...

[Yabba @YabbaYay]

@tumble_wood @Chand0___ @SpencerKSchiff @brickroad7 For GPT, not just predicting the next X, but predicting the next *word*. Focuses intelligence on concepts, rather than the world itself.

Multi-modal models will predict not just next word, but next video frame. Focuses intelligence on concepts *and* the world.

[Yabba @YabbaYay]

@SturnioloSimone @cichuck @tegmark @LukeLuk3Luke I can't speak for Max, but Robin Hanson fits this description.

Hanson seems to thinks since AI can dominate via free trade (following libertarian ethics), AI won't need to wage war against humans. Therefore, the AI domination would be ethical.

Not a widely held view.

[Yabba @YabbaYay]

@drvolts Why haven’t other states advertised themselves as safe(r) from climate change related disasters?

I understand Florida doesn’t want to advertise their problems. But why aren’t other states trying to lure retirees using climate change arguments?