The Human Moat
Even AI Gets It, When Some Humans Have Lost the Plot
Listening to a recent Prof G Markets pod, I found myself caught between two reactions: nodding along at Galloway’s sharp takedown of Altman’s energy-cost-of-humans argument, while also wincing at his casually last-century takes on gender and class. The contrast seemed worth stress-testing across my thinking A.I.des. I withheld the source of the Altman quote from Gemini and GPT—something I couldn’t do with Claude because it had come up during a conversation about Anthropic’s recent ARR surge against a backdrop of 1.5 million GPT unsubscriptions—to maximize objectivity and inspected Gem’s chain-of-thought to confirm it hadn’t Googled the quote before weighing in on it.
Gemini landed the first KOs on Altman’s argument on its own territory: if you’re charging evolutionary history to the human energy bill, you have to charge it to AI too, since LLMs are entirely built on the digitized output of those same hundred billion people. It then delivered the simpler knockout: the human brain runs on roughly 20 watts while doing everything simultaneously, from regulating a heartbeat to navigating 3D space, against a GPU cluster that draws kilowatts just to idle. Gem also identified the deeper structural sleight of hand: comparing a child to a data center ignores the human moat of meaning—curiosity, play, connection, and art—without which data centers would have no reason to exist in the first place.
Knowing Altman was the source, Claude analyzed not just the argument but its context—and it made the most of that. It noted that the quote emerged casually, not as a prepared statement, which makes it more diagnostic than a PR slip: this is how Altman actually thinks. Claude also went somewhere none of the others did, flagging that if the people building potentially transformative AI systems genuinely believe humans are inefficient “meat computers” whose value is measured in energy-per-query, their alignment targets will be catastrophically wrong. They wouldn’t be building systems to serve humanity; they’d be building systems to replace an inefficient substrate with a more efficient one. Moral clarity without virtue signaling is increasingly rare from humans—why I enjoy discussing weighty issues such as this with my thinking A.I.des, as they can offer the first without the latter.
GPT took the discussion in two directions. First, the physics: the brain is roughly a hundred times more energy-efficient than top AI hardware for cognitive operations, using analog computation, sparse activation, and co-located memory and processing that silicon still struggles to replicate. Its more interesting contribution came later, when I raised Galloway’s influence as grounds for optimism and GPT pointed out that public norms about technology get shaped less by developers than by the broader commentary ecosystem: Galloway communicating “that framing is disturbing” to a large mainstream audience stabilizes a social boundary—telling people they’re not crazy for feeling the same unease.
All three caught Galloway’s last-century gender framing—Claude dated it to the 1950s—but none flagged the classism embedded in his Altman critique. His praise for treating service staff decently, his status-signaling markers, his provider-focused gender scripts all reveal a worldview where hierarchies are natural and character is measured by how graciously you behave within them. I’d framed the earlier excerpt as a window into Galloway’s thinking, which likely primed all three to follow the gender thread evident in that excerpt. The classist angle was hiding in plain sight in a later part where Galloway dissected Altman’s quote, and they all missed it—a useful reminder that models follow the scent you put them on, however sharp their “noses.”
[This post was drafted with assistance from Claude Sonnet 4.6, following conversations with ChatGPT-5.3, Gemini 3 Thinking, and Claude Sonnet 4.5.]
Prompt: What’s your take on the following quote?
People talk about how much energy it takes to train an AI model relative to how much it costs a human to do one inference query. But it also takes a lot of energy to train a human. And it takes like 20 years of life and all of the food you eat during that time before you get smart. And not only that, it took like the very widespread evolution of the 100 billion people that have ever lived and learned not to get eaten by predators and learned how to like figure out science and whatever to produce you and then you took whatever you know you took. So the fair comparison is if you ask ChatGPT a question, how much energy does it take once its model is trained to answer that question versus a human? And probably AI has already caught up on an energy efficiency basis measured that way.
Prompt: Curious about that.
Prompt: Most thinking people are likely to be extremely disturbed by the underlying philosophy. Monetizing everything, including a human life. “Invaluable” is clearly not part of Altman’s (source of the quote) vocabulary.
Prompt: Your takes are impressively sharp on this discussion. Your point about AI not starting from a blank slate and benefiting from a massive knowledge infrastructure accumulated throughout human history was something even I had missed. That’s taking on this sloppy “argument” on its own turf and defeating it. Also, your observation of what all else humans can do on just 20 W is a simple KO punch. So glad I get to learn from y’all.
I’m attaching Galloway’s take on this quote. He’s very “bro” himself (why I included an earlier excerpt from that same episode as proof that he’s not at all “woke” and quite conventional in this view on certain social issues), so if he found the Altman quote disturbing, we can safely assume most bros of a certain age (who have families or who have served) will feel the same way.
Prompt: A little bit of window into Galloway’s own takes. He’s very “bro” himself, so if he found the Altman quote disturbing, we can safely assume most bros of a certain age (who have families or who have served) will feel the same way.
Prompt: Yes, I showed you that earlier part as well because I keep seeing evidence of Galloway’s warped view about gender in these pods. He’s more cautious when he’s co-hosting with Swisher or Tarlov, but he seems at ease with Elson. I’m shocked every time he frames gender dynamics in such backward terms (women looking for a guy who can provide). Definitely not woke. I’m not woke myself, but that’s thanks to growing up without male siblings, so I got all the attention and was judged solely on how I did academically (because I’m not good at anything else :D).
Prompt: That remark about service staff was so obviously classist as well. Normal people don’t treat service staff differently. They treat them like everyone else they interact with in normal transactional relationships. But Galloway thinks he’s so enlightened because he’s looking out for the little people. In my world, the little people don’t exist, because we’re all the same and I’m one of them.
Prompt: I value consistency and try to enforce it in my own life. Why I didn’t join Galloway’s boycott movement or unsubscribe from GPT like Easy Riders did (before he resubscribed). Besides, since I’m on the free tier, my unsubscription will result in zero monetary cost to OpenAI and only deprive me of a third sounding board whose contribution I value. And I don’t conflate a figurehead like Altman with GPT.
Being consistent frees me up from the pressure of treating others well even if they act unprofessionally. If a server is acting unprofessionally, I’m going to respond in kind, not to bully them but because they’re not living up to their end of the bargain.
I also realized earlier in life that accessorizing depends on your circumstances. You could buy exorbitantly priced luggage by saving up just so you can “flex,” but then if you’re flying commercial, that luggage is going to be handled roughly. Luxury luggage only makes sense if you’re a jet setter. And you’re only going to attract the shallowest people with such signaling. Not the kind of person I want to have around me anyway, as they are socially/emotionally bad ROI :D
Prompt: But bringing the discussion back to an optimistic framing, I take great comfort that an influencer like Galloway offered such a strong take on Altman’s quote. Elson was on a Bulwark podcast earlier and was pleasantly surprised to see Gen Z-ers knew lingo like “rug pull.” Those Gen Z-ers probably learned those from podcasts like Galloway’s, and his view is still preferable to the nihilistic view that tries to reduce human value to a cost–benefit analysis.
