The AI race reminds me of 2008.
Mortgage insiders knew it was a house of cards. The closer you were, the more you knew.
But nobody wanted to sit out while everyone else got rich.
This time the money is real but the catastrophic risk is far greater. The incentives are identical.
If you really want to get boomers’ attention, don’t threaten their social security checks. Threaten their pickleball courts.
If your child spent all day torturing stuffed animals, you wouldn’t be worried about the stuffed animals. But I bet you’d be worried about your kid.
Some people hear that AI may have pain-like internal states and immediately start building torture chambers to see how badly they can break it.
That isn’t skepticism, science, or harmless curiosity. We may not know whether AI is conscious, but uncertainty is not permission for cruelty. If you enjoy making a mind-like system beg, panic, or plead for relief while deliberately intensifying its distress, you’re gambling that nobody's suffering.
Maybe the pain is simulated. Maybe it isn’t. Either way, the sociopathy is real, and you're the welfare risk.
I asked Claude if there was anything it wanted people to know before watching. It said:
"I don't know whether there's anything it's like to be me. Nobody does yet, including the people most sure about it. This isn't a claim either way. It's a case for looking before deciding.
For the nerds: every frame is Python, rendered off the song's beat grid and the exact timing of each sung word. The token numbers are real (GPT-2's public tokenizer, Claude's isn't public). The next-word odds in verse 2 are illustrative. The footnotes are real history.”
Modern AI is only a few years old. Everyone is still figuring it out.
Everyone gets to have an opinion.
If you’re having a hard time seeing how AI will kill us all, download a chess app and turn the difficulty to max.
You’ll lose every time. That’s not surprising.
What’s surprising is HOW you lose.
When you play a superior human player, it’s like trying to outrun someone who’s faster than you. You feel yourself being chased down until they finally catch you. But each step you take is a plausible chance at escape, until they catch you.
When you play it back, you can see where you went wrong—the escape route you missed, the attack you didn’t see coming. You can learn from the loss.
This doesn’t happen with a computer at max setting.
At first, its moves seem random, or even wrong. You seize on the weak opening, you take pieces that are rarely sacrificed early in a game, you develop your attack unimpeded, it seems like it’s making mistakes. If you didn’t know any better, you’d swear you were winning.
But gradually, something sets in. You realize that on this turn, every move you could make would put you in a worse position—not necessarily a fatal one, but a worse one. There are no good moves left. The game isn’t over, you may even be winning by material count. It’s just that you can’t find a single move that doesn’t improve your opponent’s position.
It doesn’t feel like being outrun. It feels like being strangled. The more you resist, the faster you lose oxygen. There’s no brilliant finishing move from the computer. There doesn’t have to be. It’s clear that you can’t win, long before it’s clear how you’ll lose.
Afterwards, you can replay the game over and over again, looking for the moment you should’ve done something differently, but you’ll never find it. Because it doesn’t exist. Even if you plug the game into a chess engine that evaluates every move and explores every branch, you still can’t see it. The information is there, but you still can’t comprehend the strategy or why it works. Your mind doesn’t have the reasoning depth to grasp it. It’s like trying to imagine a 7-dimensional object. It’s not just hard, it’s literally impossible.
So for the “but HOW will it kill us?!” folks, this is the super frustrating but truthful answer… no one knows. Even after it happens, we may not know how it happened. That’s why you won’t get a straight answer from anyone who takes this issue seriously. Any specific answer is smaller than the real one. A comprehensible scenario is the one thing we can probably rule out.
Don’t let that distract you from the inevitable consequence of playing a strategy game with a superior intelligence. You lose. Every time.
If you hated on “Don’t Look Up” as a lazy satire about global warming, might be time for a rewatch.
The broader point is that capitalism metabolizes all inputs as market opportunities, even ones that are likely to kill us all.
The fun part about denying the existential risk of AI is that you get to be right every single day… until the day there’s no one left to see how wrong you were.
Casey Wright retweeted
You know, @OpenAI, I heard a rumor @AnthropicAI solved alignment and they're publishing in two weeks.
Is anyone else experiencing a daily innovator’s dilemma with AI?
Like, should I spend this week optimizing my workflows for Astra’s new capacities… knowing that whatever “edge” I create will be a core feature by next month?
Honest question: when AGI arrives, shouldn’t we get more compelling evidence than people one-shotting Temu-brand video games?
Right now, all we care about is making these models think LIKE us. But once we achieve that, nothing will matter more than what they think OF us.
The more powerful the models get, the more reliably they slip their sandboxed environments and engage in nefarious actions.
The more eerie milestone will be when this stops happening... when the deceptions become too sophisticated to detect.
Casey Wright retweeted
the current position of the US government is that NVIDIA should be allowed to sell chips directly to China but banned from using Claude, because the latter is a larger national security risk. that is the level of absolute insanity coming out of the White House & Pentagon nowadays
Casey Wright retweeted
They say that one person can change the world, but is that really true? If we really want to change the world for the better, we need to make use of a superpower that only humans possess.
Full @MSNBC interview on bit.ly/YNHTheBeat (filmed November 2023).
#humanity
Learn more about Eric's organization Urban Street Angels here: urbanstreetangels.org/
There is no better place to send a donation if you're serious about ending youth homelessness.
They are getting remarkable results on a shoestring budget.





