Great essay! I fully agree (and I'm an engineer and IT professional)
AI as an industry today is all about hype, greediness and power. Humanity and its (positive) future have no real place in their plans. They want money and an oligopoly that will keep challengers at the gates and ordinary people at their mercy.
AI can be useful but not in the hands of those corporations. Besides, it remains today so unreliable and there are so many negative effects in the medium term (loss of skills, loss of creativity, cognitive decline ... and what about dependency towards a handful of suppliers?). Let's proceed step by step and with reasonably sized actors, not gorillas in the cage ...
"The Democrats swept the midterms and owe their victory to the AI industry." Well, among many other reasons.
But fundamentally, this is indeed how many people feel about AI: "just don’t build the thing that will make my life a fucking nightmare." They don't care about the potential benefits if the costs to them are too high.
I'd appreciate your thoughts on AI as "an awesome learning tool". I look on AI as a poor communicator when contrasted with original sources. Original sources provide the versions that best reflect the authors' thinking. I analogize AI outputs that summarize writings largely from unspecified authors to the shadows in Plato's cave. I always feel cheated except when I have prompted for factual answers. Even then I'm wary.
Thank you John. aI should always be used with caution because it can make up stuff. But the possibilities are broader now. I don't think it can get deeper than traditional methods of learning but the breadth it enables is unprecedented, especially with agents to make interactive lessons and that kind of thing
I hope you'll dig into this sometime. I believe most people are not book learners. They learn from successes and failures. If they do remember what they read it helps if it's relatable - heroes, villains, quests, obstacles. Since most public education is designed by committee, I don't see that AI will make much headway. My thoughts. I'm interested in yours.
The Rob example is the strongest part, because it asks what the industry actually builds for him. Is there any present tense benefit you think the industry could point to honestly?
Sure. AI is a productivity enhancer for some professionals. If you know how to use it, it's also an awesome learning tool. There are many use cases but its usefulness depends on the user
Whether AI is behind Rob's higher bill depends a lot on which grid he's on. On the PJM grid, which reaches from New Jersey to Chicago, data center demand pushed up the price the grid pays to keep power plants on standby, and every home on it shares that cost: Maryland's average home bill went from $126 to $202 a month in five years. In most other states, the rise has much more to do with old wires and their upkeep. The state-by-state numbers are in here: https://marlowbrief.substack.com/p/are-data-centers-raising-your-electric
Rob's question is the whole essay in one line, and no rebrand answers it. You can't message your way out of a utility bill or a layoff notice. The industry keeps selling future-tense miracles while people pay present-tense costs, so the fix is shipping something Rob can feel this year, not a better adjective.
I'd wager a chunk of the AI industry figures they just have to build an AI so super-smart and high-IQ that it can persuade people up is down and cats are dogs. Once your gigachad super-AI is up and running, it will always have its ultra-perfect hyperintelligent arguments that are not at all affected by pesky things like "material reality" and "experience".
Low blows aside, I suspect the AI industry will at least try having its latest AIs handle its own image problem within the next two years. The machines might not even do worse than the kind of nerds doing the job now.
Alberto, I recently finished “Against the Machine” by Paul Kingsnorth. You are both among a handful of substacks I eagerly look forward to reading each week. I’m starting to think it is fantastic for AI to have a PR problem, for the sake of our souls.
I am not sold on the election claim at the top. Alberto even says in the replies that AI was not one of the top three reasons, so calling this a verdict on AI feels like a stretch. What Rob pays for power and what Natalie pays in rent are the real story, and tying them to a midterm sweep lets the politicians who set those energy and housing rules hide behind a tech story. The industry has a real problem, but making the election about AI probably makes it harder to fix, not easier.
It's not the main reason why the Dems won that's for sure but insofar as it is a reason, people made it clear they are vehemently against it. That's why this post was worth writing. I don't think AI is the most pressing issue in the world or America, contrary to many people in Silicon Valley, but it's the issue this newsletter is about!
Thanks Alberto, this is very helpful just now. We want to deflate the bubble not pop it. We don’t want another emergency that allows the oligarchs to make their killing and leave us holding the bag.
From here on it’s not about ‘the messaging’ it’s about an actual negotiation with the rest of the country represented.
Great essay! I fully agree (and I'm an engineer and IT professional)
AI as an industry today is all about hype, greediness and power. Humanity and its (positive) future have no real place in their plans. They want money and an oligopoly that will keep challengers at the gates and ordinary people at their mercy.
AI can be useful but not in the hands of those corporations. Besides, it remains today so unreliable and there are so many negative effects in the medium term (loss of skills, loss of creativity, cognitive decline ... and what about dependency towards a handful of suppliers?). Let's proceed step by step and with reasonably sized actors, not gorillas in the cage ...
"The Democrats swept the midterms and owe their victory to the AI industry." Well, among many other reasons.
But fundamentally, this is indeed how many people feel about AI: "just don’t build the thing that will make my life a fucking nightmare." They don't care about the potential benefits if the costs to them are too high.
Haha yes. I don't even think that's one of the top three. But it's more punchy like that.
Clickbait! :)
The Medium is the Massage.
This essay is much appreciated.
I'd appreciate your thoughts on AI as "an awesome learning tool". I look on AI as a poor communicator when contrasted with original sources. Original sources provide the versions that best reflect the authors' thinking. I analogize AI outputs that summarize writings largely from unspecified authors to the shadows in Plato's cave. I always feel cheated except when I have prompted for factual answers. Even then I'm wary.
Respect,
Thank you John. aI should always be used with caution because it can make up stuff. But the possibilities are broader now. I don't think it can get deeper than traditional methods of learning but the breadth it enables is unprecedented, especially with agents to make interactive lessons and that kind of thing
I hope you'll dig into this sometime. I believe most people are not book learners. They learn from successes and failures. If they do remember what they read it helps if it's relatable - heroes, villains, quests, obstacles. Since most public education is designed by committee, I don't see that AI will make much headway. My thoughts. I'm interested in yours.
The Rob example is the strongest part, because it asks what the industry actually builds for him. Is there any present tense benefit you think the industry could point to honestly?
Sure. AI is a productivity enhancer for some professionals. If you know how to use it, it's also an awesome learning tool. There are many use cases but its usefulness depends on the user
Whether AI is behind Rob's higher bill depends a lot on which grid he's on. On the PJM grid, which reaches from New Jersey to Chicago, data center demand pushed up the price the grid pays to keep power plants on standby, and every home on it shares that cost: Maryland's average home bill went from $126 to $202 a month in five years. In most other states, the rise has much more to do with old wires and their upkeep. The state-by-state numbers are in here: https://marlowbrief.substack.com/p/are-data-centers-raising-your-electric
Rob's question is the whole essay in one line, and no rebrand answers it. You can't message your way out of a utility bill or a layoff notice. The industry keeps selling future-tense miracles while people pay present-tense costs, so the fix is shipping something Rob can feel this year, not a better adjective.
I'd wager a chunk of the AI industry figures they just have to build an AI so super-smart and high-IQ that it can persuade people up is down and cats are dogs. Once your gigachad super-AI is up and running, it will always have its ultra-perfect hyperintelligent arguments that are not at all affected by pesky things like "material reality" and "experience".
Low blows aside, I suspect the AI industry will at least try having its latest AIs handle its own image problem within the next two years. The machines might not even do worse than the kind of nerds doing the job now.
Alberto, I recently finished “Against the Machine” by Paul Kingsnorth. You are both among a handful of substacks I eagerly look forward to reading each week. I’m starting to think it is fantastic for AI to have a PR problem, for the sake of our souls.
I am not sold on the election claim at the top. Alberto even says in the replies that AI was not one of the top three reasons, so calling this a verdict on AI feels like a stretch. What Rob pays for power and what Natalie pays in rent are the real story, and tying them to a midterm sweep lets the politicians who set those energy and housing rules hide behind a tech story. The industry has a real problem, but making the election about AI probably makes it harder to fix, not easier.
It's not the main reason why the Dems won that's for sure but insofar as it is a reason, people made it clear they are vehemently against it. That's why this post was worth writing. I don't think AI is the most pressing issue in the world or America, contrary to many people in Silicon Valley, but it's the issue this newsletter is about!
has anyone in that industry tried answering in the present tense yet
Thanks Alberto, this is very helpful just now. We want to deflate the bubble not pop it. We don’t want another emergency that allows the oligarchs to make their killing and leave us holding the bag.
From here on it’s not about ‘the messaging’ it’s about an actual negotiation with the rest of the country represented.