They could make a blimp to carry it. It'd have to be a really big blimp.

A useful rule in many math courses is that everything stated is important in the solution. If there's anything stated you haven't used, you're making a mistake. Adding irrelevant things breaks that rule, making finding the solution less constrained, taking more time to search for the solution. And if the student knows the rule but doesn't know it can be broken, suddenly everything is hopeless because the rule forces the irrelevant thing to be used in the solution.

Suno AI is way better at performing music than I am. We're similar in composing music, but AI is much faster and seems unable to listen to my melodies, so I usually let Suno do its own melodies as well as its own performance. I supply lyrics and it's often adequate at using them.

Sometimes Bing Image Creator can give me a picture I want. Sometimes it's easier to draw it by hand. Sometimes the images initially look good, but on further inspection they've got some horrible references hidden in them.

ChatGPT has been useful for conversations ... it's like a ridiculously well-read professor who's high on pot, he knows everything but he doesn't think anything through. It's led me to some interesting rabbit holes.

Work keeps encouraging me to use AI for code, but so far it looks like a very fast and very junior developer, and I haven't motivated myself to waste my time on that so far. A couple times I've successfully asked it for interfaces for doing X in HTML. I had a meandering conversation with ChatGPT on optimizing if-else chains and together we found one I hadn't seen before (sum 0 or (1<<i) for independent conditions i=0..n, use machine instruction to find least set bit in sum, switch on position of least set bit).

I can buy that it's not influenced by any classical communication between the stations between when one photon is sensed and the other is sensed, but being fundamentally random is a stronger claim.

I just asked Bing Image Creator to portray a pickleball match with darth vader and elsa vs mickey mouse and shrek, and it did. Although it had darth vader's head on elsa's body, and a volleyball, and they were playing at night, with court lines on the sky, and mickey mouse had a curvy figure, and so forth.

I've seen people asking Amazon to label whether products were made in America, so they could buy American. Labeling products not made in America with the tariffs that have been applied because they're not made in America seem like the same thing. Not to mention that Trump believes tariffs are beautiful. Why would he object to spelling out which items have tariffs and how much they are?

We've been using computers to help with math since the invention of computers. For example the four color theorem from 1976 was famous for being too big for a human to verify manually. Coming up with a theorem involves spotting a likely truth, finding an explanation, and verifying the explanation is correct, and computers are used for all of those.

We're software developers, yah? You monitor how your software performs in real life. You see what rabbit holes it gets stuck in, and you rearrange the system or design features so it doesn't get stuck. It's constant firefighting, trying to get the software to flow in reality rather than getting all tripped up. That's what software welfare is. If AI gets to the point where having a bank account would let function better, you can bet we'll give it the right to have personal bank accounts.

Study Abroad programs are notoriously party schools. Even if the school itself is decent, the study abroad portion makes it easy to not get an academic education.

Answers, $0.01. Answers with thought, $1. Correct answers, $100. Snide remarks are still free.

(There was a poster something like this in the 1980s, but I can't find any reference to it. I don't remember how the saying originally went.)

The real win of tax software is downloading all the data rather than typing it in. I wrote my congresscritters at least once saying that the free tax software is great and should do this, but also how I wanted those TurboTax parasites to die. Having the IRS publish tax-filing software that does the downloading is such an obvious win for the people. And, so, obviously something Trump will shut down, since he shuts down everything by-the-people-for-the-people about the government.

Incidentally, the right way to download the data is not to give TurboTax the username and password to your bank accounts, which would enable it to drain your bank accounts if it wanted to. The right way is for the bank to give you a specialized username/password for just downloading the forms with the data, which you then give to the tax filing software. I had a bank account that used to do it the right way, but they got bought and merged, so this year it required giving the whole-account username and password.

You overestimate the liberal arts content of a four-year compSci degree.

Interesting! I responded by getting a new computer and reimaging it to Linux ... I hadn't considered plain switching the old Windows computer to Linux.

I don't think it's practical. Anytime I change my home computer, I want the old and new one working at the same time for awhile while I iron out the workflow with the new one. Switching from Windows to Linux is a bigger-than-normal change in workflow.

A software coding agent should do the job of a system software engineer. It'd be a co-worker not a tool. If it's good and fast enough you can fire all the human workers.

For example: the project at work precreates containers at a constant rate. Precreated containers depend on the state of the system, so they age out after awhile as the system changes. If the rate were dynamic based on past behavior, it'd use less resources most of the time and avoid running out during emergencies. There's also some control code that does ... something ... when there are creation failures. The AI agent ought to be able to read the existing code, implement dynamic precreation, its code should be terse and readable, it should be able to explain and defend its code when we code review it, it should obey the existing system conventions, it shouldn't introduce more bugs than a human would. It should be able to code review other people's and agent's code. If it's good, it should look at the system and point out that dynamic precreation should be done in the first place.

When I see such code changes proposed by AI agents, I'll believe the hype, but not until then. Sort of like a self-driving car is one that can substitute for a human driver.

Most of those problems could and should be solved in Antarctica (or Texas). Cheaper and faster. Isolation, recyling, minimizing resupplies, growing crops. Living in microgravity can't be solved on earth, but the solution to that one is to have a rotating habitat and not live in microgravity.

Once Musk's starships are going strong, hook twelve together in a torus and spin them and you've got a minimal rotating habitat. But why? What problems can be solved on that that you can't solve cheaper and faster on earth? Living in microgravity would no longer be a problem needing solving. Mining and manufacturing in microgravity are still things you can only do in space.