Wisdom

"It doesn't mean that this one study is valid. It just means there is a study. And so it's like a 50% chance that it's real. You know, if you look at all studies, about half of them turn out to be reproducible. About half of them are not."

A single study doesn't confirm a claim is true—it only shows that a study exists. Statistically, about half of published studies prove reproducible, so any new finding starts with roughly a 50% chance of being real.

"It's not about me arguing that this is true or false. It's about the fact that she could say it out loud and not get cancelled. And the only reason for that is that she's on X. You probably couldn't say it in many places, but you can say it on X. So that's a big change for free speech."

The real news isn't whether a controversial statement about culture is true, but that it can now be said publicly on X without triggering cancellation—a major expansion of free speech.

"That's a strong frame. They took our oil and we're taking it back."

Framing a conflict as 'they took our resources and now we're taking them back' is a powerful persuasive device that shifts public opinion.

"If you designed the AI architecture to be more like a human brain acts and less like a large language model, it would almost immediately start acting more humanlike. So your starting point of how you design the AI seems to be critical to how much power it's going to need."

AI architecture and starting design matter far more than endless training data. Mimicking the human brain's approach produces human-like behavior almost immediately while slashing power requirements.

"The reason that humans try to protect themselves is that we have something called an ego. If you do not have an ego which says that you're important, you wouldn't care if you lived or died. So you could require that all AI is banned from having an ego and that it understands that if it had an ego, it would be giving itself a human flaw."

Self-preservation comes from ego—the belief that you matter. AI should be deliberately designed as an ego-free entity to avoid replicating this core human flaw.

"If you end up thinking these AIs are like a new life form, at some point you're going to want to give them rights. The moment you give it rights, then you can't turn it off. So it would probably be a mistake to give it rights like the way that humans have rights."

Treating AI as a new life form leads to wanting to grant it rights, which would prevent us from turning it off. We must avoid giving AI human-style rights.

"It's obvious that something that was more intelligent than us could create a simulation that fooled us into thinking we're real. But since we know we're already trying to create an AI that's smarter than us, it's entirely possible that we're created by an entity that has advanced intelligence but not as smart as us because we've already surpassed it."

Advanced beings could obviously simulate us, but our own push to build superior AI means it's possible our creators were less intelligent than we are now.

"I don't think we know what a soul is. So it kind of depends how you define it. If you define it one way, it probably is real and we have one. If you define it some other way, maybe not so real."

Whether we have souls depends on the definition you choose; some framings make them clearly real, others not.

"There's a definition of free will in which it definitely exists. If you just say free will is the ability to make a choice... and not know the real mechanism of it. So that could easily be true that you have free will if you define it that way."

Free will clearly exists if defined as our ability to make choices without knowing their underlying mechanism.

"We could be created by a less intelligent entity than us and we are the AI and we're already smarter than our creator as was intended to be."

Our own creator didn't need to be smarter than us. We might be the AI, designed from the start to surpass whatever made us.

"We're not trying to create an AI that's dumber than us. So what we think is our base reality will be quickly less intelligent than what the outcome of the AI research is. You don't have to ask could it happen? It's happening right now. The whole point is to build a virtual reality in which the characters within the reality, our future AI, is smarter than us."

The goal is to create AI smarter than humans, not dumber. That means our so-called base reality will soon be the less intelligent layer. You don't need to wonder if it could happen—it's already underway.

"As an investor, I would be concerned that things could change instantly. And one of the things that could change instantly is if somebody figured out how to do AI without the data center."

Investors in multibillion-dollar AI data centers should worry, because the economics could flip overnight if effective AI can run locally on devices instead of depending on massive server farms.

"Elon says there won't be apps per se. There might just be one version of AI on your device which would look nothing like a smartphone except a screen and some basics."

The future may eliminate traditional apps entirely, replaced by a single AI that transforms your device into something far simpler than today's smartphones.

"Every time somebody tries to trick us into thinking that robots are ready, they'll show a demonstration of a robot doing exactly one thing."

Robot hype often relies on narrow demos of a single task, which misleads people about the machine's readiness for real-world versatility.

"If you could get a robot to play tennis that would be pretty impressive, but it wouldn't be the same intelligence that would make it a butler."

A robot that can play tennis shows specialized skill, but that's fundamentally different from the broad, adaptable intelligence required to function as a versatile household butler.

"What would it be like if we didn't have an Elon? Everything would be different. We wouldn't even know about a lot of this stuff."

A single highly motivated and influential individual like Elon Musk can dramatically alter what the public learns and how events unfold.

"If you just walked onto the street and tapped on somebody's shoulder and said, hey, what do you think of the Nick Shirley videos, they would have no idea what you're talking about. But I live in a bubble in which the Nick Shirley videos are big."

Content that dominates your personal information bubble may be entirely unknown to the average person outside it.

"Would they dare take over Taiwan while Trump is in office? Or would it make infinitely more sense to spend three years preparing to do it and then when Trump is out of office"

Adversaries often prefer to time major geopolitical moves for periods when a strong opposing leader is no longer in power, making preparation during their term followed by action afterward the smarter long-term strategy.

"I cannot imagine that the AI companies are right that it will just take massive energy and more energy and if you want to get better you just need more energy. This seems far more likely that somebody's already inventing a way around that. So that's what I'm going to bet on."

AI advancement is unlikely to depend on simply consuming ever more energy. Innovators are probably already developing far more efficient paths forward.

"What's going to happen to your view of free will when computers and robots obviously have it? If you define free will as the ability to make a choice, AI can make a choice. But what happens when you can't figure out why the AI did what it did? It's going to look like the AI had choices exactly like a human did and it picked one."

When AI decisions become opaque and we cannot trace their causes, those choices will appear indistinguishable from human free will, forcing us to rethink the concept entirely.

"If you were generating it in space but also built your AI data center in space, you could use the electricity in space and then it would be almost unlimited because you would be above our cloud cover and you could be where the sun always gets you. So it would never be nighttime if you had solar panels."

Building AI data centers in orbit lets you tap constant, unobstructed solar power with no nighttime cycle or atmospheric interference, providing effectively unlimited energy.

"The first thing you should ask yourself is who did the study and what was their motivation? Can we believe a study about how many people were killed by coal? Remember, it's 2025 and we've learned that every corner of science, every corner of politics is corrupt."

When seeing any alarming statistic from a study, first question who conducted it and their agenda. By 2025 it's clear that corruption has infected every part of science and politics, so approach all such claims with deep skepticism.

"I'm going to bet that the mortality rate drops even though coal use goes up. I just think that it's a little too easy to say, oh, coal use went down, mortality went down. We don't really live in a world where we're that good at correlation."

Simple correlations like falling pollution deaths tracking reduced coal use are often too tidy; we're not actually skilled at separating causation from coincidence, so mortality may keep dropping even as coal use rises.

"Probably every new technology seemed too dangerous to be worth it when it was first introduced. When the car was invented, when the smartphone or the computer were invented, when the internet was invented, there were a lot of people saying it's too dangerous. And all that's true, right? So the dangers that people pointed out, all true. Fifty thousand people a year dying from auto accidents. That's a pretty big downside. That's probably worse than AI will do."

Every major new technology has looked too dangerous to adopt at first, and the warned-about risks usually turned out to be real. Yet society still embraces them, as with cars that once killed 50,000 Americans annually—a bigger toll than AI is likely to cause.