I'd use it to get a degree in an AI proof industry, like dental hygienist. Nursing kind of sucks these days because of AI and healthcare business. Other AI proof industries might be aromatherapy and massage. Can't even go into a nice peaceful job of driving a garbage truck, robotics are going to end growth in that industry in a few years.

It's an interesting formula MS has though. Age + years of employment >= 70. But I'm not sure how much they are actually paying. Probably a formula of years of employment each get a fractional multiplier on base salary.

I propose prison for anyone who has over $1B in assets. Every month they spend time in a cell deciding how badly they want to keep their money. At the end of the month, they can decide if having a modest $900M is all one person really needs. Why would I go after someone's money, they "earned" it right? It's simple, it complicates a society that operates under the assumption of 1-person-1-vote. That at some fundamental level we are all equal in our representation and voice. That a shopclerk has as much right to their opinion as a the franchise owner of that shop.

Anthropic has real products and real revenue. What people overlook is that anyone can operate an AI company. And there is no secret sauce to making the models. And there are thousands of good quality and free models. And most importantly, a frontier model is obsolete in a matter of months right now. So even if Opus and Fable are the best in their class (debatable!), they won't be next year. So it's a constant churn to build new models. And while you can reuse the training and processes for your model, the model itself is not really something you can arbitrarily extend (nor would you want to), so there is a lot of starting from scratch in this industry.

It's low risk if you were in on it long before the IPO. Now that it is announced, it is just part of the grab back that crappy financial managers stuff into the middle class investment portfolios.

The entire surface of the Earth should get transformed into data centers, and we humans can survive underground. That's the future ASI and the tech billionaires promise. The billionaires get to live in towering air cities, floating ships, or orbitals. Which are nearly practical today using personal drones for transportation.
While AI could have lived underground and humans on top, why would a billionaire want to look at poor people or let them have access to solar energy or valuable real estate?

While it's true every issue has its own dimension, even if those dimensions are linearly independent, they're not necessarily orthogonal. Independent basis vectors can still have a non-zero dot product—in other words, issues which are distinct from each other can align to some degree.

If you stopped irrigating crops in the US desert southwest, there would be no water issues at all, either for households or for industry.

I'll repeat, it is agriculture that consumes meaningful amounts of water, not homes and businesses. Farms by water by the acre foot - an acre of land flooded to a depth of one foot. Water is not abundant on the scale of "farms buying it by the acre foot", but it absolutely is on the scale of "what you need for your house".

How is AI different? I feel like you didn't finish your thought.

Yes, the power hammer and the factory did eliminate most blacksmith jobs. And the tractor (and its many machinery cousins) did reduce the need for farm workers by 97%. While blacksmiths and farmers are still needed to a degree, it's still valid to say that their jobs were replaced by technology.

On the other hand, there are new jobs today, like farm combine manufacturing workers, and equipment repair jobs, and equipment software developers, and factory designers, and factory builders. These jobs didn't exist before the technologies were invented. You might argue that not as many new jobs were created, as were eliminated, but employment statistics beg to differ. We are still, despite all the hand-waving, at full employment, at least in the West.

AI is *not* different, in any way, other than the specific thing that is being automated. And by that measure, every automation technology in history, automated a specific thing that hadn't been automated before.

I think it best to let the hardware deal with error correction. It is way better positioned in the stack to judge and handle these sorts of failures at least cost. At best all you need are simple checksums to alert the operator to hardware failure.

Freshwater is cheap for a reason: it's abundant.

Uh, no. Fresh water is abundant in some places and scarce in others. The reason it's cheap in the US in places where it's scarce like Arizona and Southern California is that the government subsidizes it, as well as drawing more water from aquifers than can be replenished in time. In a lot of places in the USA, water consumption is not sustainable. And yes, agriculture is the main culprit. Exporting oranges from California is just insane; you're sending out crates of 90% water from a desert.

I did get a degree, I do know what university is, and how it works.

Some kinds of college classes do require "homework." I took college level music classes, for example, which required 15 hours or so per week of practice. If I didn't do my practice, the results were evident at my next meeting with my instructor.

For a different type of work, like, say, a research project. Sure, a student could get AI to do the project. But then a meeting with the professor or GA, or a presentation to the class, or written answers to questions done during class time, would quickly expose that the student did not in fact do the work. It would take some effort to design the evaluation, depending on the specifics, but it's not impossible to know whether the student did the work.

But many college classes assign homework just because the professor or school thinks homework should be given. Take College Algebra. The purpose of homework should be practice. Grading of the homework shouldn't be necessary. If the student can learn the subject well enough to pass a test without practice, why not! I actually did this, I tested out of College Algebra; by taking a single proctored test, I got credit for College Algebra, without doing the work. Nobody graded my homework, because I didn't do any homework.

Even at the college level, the purpose of homework needs to be re-thought.

You take the panels off? Oh no, you have to use an electric wrench to unbolt some stuff after pulling a disconnect lever on the inverter. Once every 30 years.

THE HORROR.

That's exactly the calculation. Use redundant storage in-flight, and make LRU swaps very quick on the ground. If you catch a failure in-flight, you have a tech with a replacement LRU waiting for the plane when it lands (for various values of "where" - obviously you don't have techs at every airport).

It's really not that different from anticipating disk failure in a server by having online redundancy, and a service callout to replace the part. Especially if you can quickly refurbish and recertify the LRU and put it right back into the replacement inventory.

This is strange. SSDs are supposed to have scrubbers that periodically scan for this and rewrite leaky blocks internally.

What part of your question points to something that is different *this time*?

AI also takes the "menial" work of, say, coding. To me, this is very definitely drudge work. Why should I personally have to solve the problem of, for example, how to decrypt an authentication token, when it's a solved problem? When I write such code, I'm not doing something new, I'm just repeating a process that has been done many times by others. I'm doing it again because it hasn't been done in *this* software before (whatever I'm working on). This is drudge work. I just want the software to authenticate. Until AI came along, I had no choice but to figure it all out again, for myself, because I didn't have the experience of those who had likewise figured it out.

AI still can't decide what software *should* do. It can only implement what it is told to do. Sure, you can tell it to just build an app and make the decisions itself. But would anybody actually want to use such software? Would it do anything useful? I doubt it.

The idea that "this time is different" gives way too much credence to the AI hype machine. The AI evangelists say that AI will soon make all white-collar jobs history, and so do the AI doomers. I don't think so. I think AI will take some jobs, and make others, just like every other technology in the history of man.