26 Comments
Sep 10·edited Sep 10Liked by Gary Marcus

I asked Anthropic's Claude to discuss AI policy in the style of Donald Trump: "Folks, we're gonna have the best AI, believe me. It's gonna be so smart, your head will spin. But we gotta be careful, because some very bad people, some very nasty people, they want to use AI against us. Not good! I know more about AI than anybody. The fake news media, they say "Oh, Trump doesn't understand technology." Wrong! I have a very good brain and I've said a lot of things. Our AI is going to be so powerful, so incredible. China and Russia, they're trying to catch up, but they can't do it folks. We're going to win the AI race so much, you may even get tired of winning. But let me tell you, we need to put America First in AI. No more of these terrible deals where other countries steal our technology. We're going to have great AI jobs, the best AI jobs. Coal miners? They're going to be coding robots. It's going to be beautiful. And security? Forget about it. Our AI is going to be like a giant wall, protecting us from all the bad hombres trying to hack us. It's going to be a cyber wall, and guess what? The robots are going to pay for it! So remember, when it comes to AI, it's very simple: Make America's Algorithms Great Again! Thank you, thank you very much."

Expand full comment

My guess is that they won't mention AI, because they know nothing of any substance about it. And that would be true pretty much of the entire Congress.

Expand full comment

Kamala Harris literally wrote the executive order on AI, so her policy is pretty clear, to me. That's why I supported her in the first place. She's the only one who is taking risk seriously.

Expand full comment

I predict Trump will most likely sound like he’s predicting one token at the time…

All jokes aside, I have little faith in AI even being mentioned once.

Expand full comment

But unlike ChatGPT, Trump always picks the best token!

Expand full comment

Trump is a token. Just a sad, old, corrupted one.

Expand full comment

Actually what they SHOULD mention is climate change and other environmental issues (biodiversity loss; sea level rise; ocean warming, acidification, and deoxygenation; etc.) but I doubt that will be debated either — aside from, possibly a mention of Biden’s legislation. This is no less historically unprecedented and urgent for humanity than AI. E.g., you won’t care much about AI if you’re starving because of crop failure.

Expand full comment

And the saddest part of it is that at this very moment X is flooded by either fake propaganda (Taylor Swift supporting DJT) or stupid images / movies / memes of Trump and Harris kissing, having a baby ... or Musk as a super hero or a gladiator fixing everything and "eradicating the woke virus". Human stupidity is growing faster than artificial intelligence that's for sure. But it's difficult to regulate ...

Expand full comment

We certainly want government to be paying attention to AI but there is absolutely no way that Harris or Trump could say anything useful about AI in a TV debate. Quite the opposite. We know that anything they say will be terrible.

Expand full comment

Cats and dogs dominated sadly. No other debates planned. What next?

Expand full comment

Maybe replacing both candidates with ChatGPT would’ve been a huge improvement? 😁

Expand full comment

One candidate probably does not know how to spell it

Expand full comment

Kamala just mentioned AI and quantum computing! :-)

Expand full comment

The politicians are not going to discuss the failures of AI because their campaigns are funded by the leaders of trillion dollar information companies (Alphabet, Amazon, Apple, Nvidia, Microsoft) that have caused the problems you cite and worse. Their technology has failed society but they have profited obscenely by the failure.

The symptom of the failure is universal reliance on algorithms (search or LLM) that use a text box UI. A text box limits users to the vocabulary they know. The social harm is caused by the most educated 10% with vocabularies of 20,000 to 50,000 words being satisfied living in a knowledge bubble while billions are forced into a knowledge apartheid by their limited vocabulary.

Gary has described a technology that provides a promptless interface to access knowledge without needing words. This technology is called semantic AI.

The 2% of families that own these companied are not users who need a promptless interface to discover concepts they don't know exist. But billions of internet users from children through high school and every unschooled child or language limited internet user does.

Semantic AI generates the LLM prompt which prevents most if not all the serious ills that need to be addressed.

This video https://bit.ly/3WuGyxE (1 m) shows how semantic AI works with generative AI. Of course the most educated could learn something new also.

The AICYC project is a nonprofit public open source effort. Learn more at www.aicyc.org

Expand full comment

Thank you for sharing this -- unfortunately it feels like there's a limited few that understand the magnitude of change AI will cause. Technology somehow still isn't considered to be a pivotal issue in policy alongside other important issues like climate, immigration and foreign policy -- despite the fact that it one of the most powerful forces at play in both domestic and international society.

Expand full comment

They won’t bring up AI…

Expand full comment

nothing substantial will come from the debate tonight. They will not be the force that can regulate the A.I. Only the people standing up and boycotting or talking to there state attorney general can do something. It has to start at the local level. The presidential candidates are to neutered.

Expand full comment