Hearthstone is absolutely swarming with bots that beat humans regularly.

Which humans? Many people are simply not that good at Hearthstone. Are the bots regularly achieving high legend ranks?

The team thing is killing us right now. Everyone is just kind of silo'd with their agents and they present these entire complex units of work with no input from the rest of the team, and by then it's too late to fix.

excellent observation.

Yeah people push the models to the limits of what they are capable of almost instantly.

> I think some people will make a lot of money and more people will lose a lot of money.

---

Yeah, this is exactly right. The numbers look impossible because they _are_ impossible. The _vast majority_ of AI spend is wasted and the vast majority of AI companies are going to go bankrupt. There are _still_ going to be unimaginable amounts of wealth earned by the survivors.

>The _vast majority_ of AI spend is wasted

I don't think this part is correct. It is spent on things. There will be a quantity that will go training models that do not contribute research data or active use that you could say that the resources were wasted. If the money is spent on infrastructure or the resulting models found active use (or provided research insights) then I don't think you can say that the money has been wasted.

I think a large proportion of the advances in AI that have come in the last year have come from being able to see the models from the previous year in action.

That does not guarantee that money will be returned to investors, Just that it wasn't wasted.

> I think what's lost on the general public who mainly thinks AI is ChatGPT,

There are so many people who have never used Claude Code or Codex who think that Google's AI results are representative of what AI can do.

I formally proved an important 2025 CS paper (which I did not write) in Lean last month, it took like 2 or 3 weeks of intermittently poking at claude to keep going. AFAIK, it is the first rust-like borrow checker completely formally proven in Lean.

I'm currently using it as a basis for building a systems language with Rust's memory guarantees, but with new features like generators and co-routines, and effects instead of colored functions. The fact that the borrow checker's properties are formally proven means I can trust more of what Claude is doing than you normally would be able to do.

I have been very tempted to but don't know if that is literally a crime or not, lol.

Driving through the rockies is _terrifying_. It's not just this one road.

I remember driving several scary roads out in Colorado. The 210 out to Snowbird in a blizzard and a few others legit made my palms sweaty and I wasn't even driving. As a great plains Midwestern kid, seeing my buddy (who was well familiar with the roads) drive as well as he did in those conditions at least helped a little.

It just reinforced I didn't have the constitution live out in CO and drive those roads in those conditions.

oh come on the road to snowbird is chill

Driving the Trans-Canada through BC was absolutely terrifying in the 80's; today it's a completely different experience. Twinning it took 25 years (1979 to 2014).

It's also a bit dizzying. Really easy to just get lost in the spectacle of the mountains.

I love the idea, but the moon has too much gravity and moves too fast, and the missiles too slow, which makes it impossible to aim. There's a fine line between complexity and chaos and this is on the wrong side of it right now.

There was a long period of time that people engaged in _accidental_ agriculture, where they collected seeds and foods to eat and sort of accidentally spread them around along migration pathways, so they probably had "fields" of edible plants around to some extent, but they probably wouldn't show up in the archeology.