On the plus side, once the AI bubble bursts and 75% of the AI companies go under, there's likely to be a lot of electrical capacity (and GPUs and RAM) suddenly available at fire-sale prices, at least for a while.

I don't think that's accurate -- the only people who "prefer to use coal" are in the Trump administration. China, like the rest of the rational world, prefers to use whatever energy source is cheapest and most effective, which might be coal in some situations, or it might be solar, or nuclear, or hydro, or something else.

She should have said "programmed" rather than anthropomorphizing it, but other than that, she's correct -- that is, in fact, how it is programmed to behave.

It depends on the density of the neighborhood. The preferred use-case for drones is "neighborhoods" where the houses are few and far apart from each other, making ground delivery tedious and making the distance between the drone and the nearest set of ears larger.

Okay, here's one field: in the last four weeks, Claude Code has detected and diagnosed 91 genuine bugs in the open-source library I maintain. That's 91 bugs that likely would have remained unfixed indefinitely, unless/until I (or a user) happened to stumble across a resulting runtime misbehavior and then laboriously worked our way backwards to pinpoint the underlying software defect. I'd estimate probably 150 man-hours were saved, right there.

Is finding, reporting, and fixing latent bugs in C or C++ code a practical purpose? Because they're doing a damn good job of that.

If you have access to its command-input interface, either you own the system and are expected to be able to manipulate it, or you've somehow obtained unauthorized access, in which case it has a security problem, and it would be an equally serious problem for a non-AI system.

You could mine Bitcoin, I suppose, but the obvious thing to do would be charge up your EV. Energy storage on wheels!

n/t

(And yeah, it was a fun degree. Just a BA :) Decided to go back to school for a second degree on a topic that had long been of interest to me)

DNN-based, like nearly all modern AI. Not Transformers, as far as I'm aware.

Explain how this doesn't count as reasoning. Or this. To name just a couple examples.

Yes, they work by fuzzy logical reasoning. That is literally how neural networks, including the FFNs in Transformers, work. Every neuron is a fuzzy classifier that divides a superposition of questions formed by its input field by a fuzzy hyperplane, "answering" the superposition with an answer ranging from yes to no to anything in-between. Since the answers to each layer form the inputs to the next layer, the effective questions form grow with increasing complexity as network depth grows. Transformers works by combining DNNs with latent states (works on processing concepts, not raw data, with each FFN detecting concepts in their input and encoding resultant concepts into their output) and an attention mechanism (the FFNs of a given layer can choose what information they "want to look at" in the next FFN).

This reminds me of the time in elementary school when my half-informed friend insisted that the only operation an Intel 8086 chip was capable of was adding 1 and 1 together. I'm pretty sure someone had tried to explain to him that at a fundamental level, CPUs are based on repeated applications of binary logic, but the lesson he took from that was that the Intel 8086 chip in particular was horribly crippled and could not do anything useful.

The "LLMs are just predicting the next word" meme is similar. It was largely true five years ago, and there's still a little bit of truth to it, but 2026-era AIs are much more complex and elaborate than that, in the same way that a 80486 is not "just a one-bit adder".

Nerd-snobbery is the funniest kind of snobbery :)

I'd be surprised if they weren't using AIs to review the explosion of AI-written code. Having AI #2 review the work of AI #1 seems like the obvious thing to do (if you're trying to avoid paying a human being to do it).