I think the end state of all this looks like game cartridges. If people could buy the weights of a frontier quality model in a high density, high speed ROM that ran locally plugged into an PCI-E or M.2 slot, all you would need then is a reasonably fast tensor processor with a little RAM for context. This is possible, and there is even a company (Taalas) with an early product. They have an online demo that is crazy fast.

You would buy one and use it for some time, probably years, and then discard it when the value of some newer version makes sense to you.

Not far from the truth.

The big implication from this is that making these inference engines isn't difficult and they can't be protected. That's the doom of all of AI money: the product is getting commoditized; you'll buy a Claude/ChatGPT/whatever in a box from Amazon in a few years.

It's taking shape: Basic life may actually be quite common. Naked apes typing on keyboards on a digital network they built themselves not so much.

The rare earth and rare advanced intelligent life theories just got some extra weight.

By the time Slashdot started using id's, I thought it had already become too crowded to have a useful conversation, so I didn't bother to register. I don't even remember why I eventually broke down and registered one.

I remember my first post was to a discussion of whether it was better for Linux distros to conform to a set of committee defined standards, or simply to allow the market to define de facto standards. Of course, that was posted under my real name, as was usually the custom at the time. I remember thinking that Anonymous Coward guy sure posted a lot of comments.

These days, I don't think I'd post anything anywhere under my real name. I don't think there's even an archive of those original Slashdot articles and comments.

But by the time you get that, she'll be over 50, and not at all as appealing.

What you don't understand is the Python is often used as a method of invoking libraries that are written in more efficient languages. And for the layer that it handles it doesn't introduce unacceptable inefficiencies. E.g., you wouldn't want to do ray tracing in Python, but it's fine for calling a library that does that.

I'm quite sure quantum computers are valid. Whether they're useful is another question. I'll agree that it's not clear that general purpose quantum computers will ever be useful. (I won't agree that it's clear they never will be useful.)

OTOH, specialized quantum computers are already useful. DWave sells one design.

There really are good use cases for fission reactors. Solar can't handle everything, even combined with wind and the grid. (But close.)

E,g,, I think that the case for fission reactors on the moon is sound, though any particular implementation may be quite questionable.

... in Europe is roughly 5 degrees centigrade above worst case scenarios projected for the year 2050 back in 2016. Germany will likely crack the 40 degree mark in multiple locations at the end of this week. Once again a new heat record. I personally expect this to only get more intense in the next years until perhaps the gulf stream completely shuts down.

These are cascading effects kicking in and ramping up. It wouldn't stop if the planet went net-zero carbon tomorrow. So we're pretty f*cked, as predicted ever since 1970. I'm curious how hard though. Guess we'll find out soon.

... in some parts, contains bucketloads of over-the-top excess trivia in others and has sections that are flat-out provably false. If the sections chiefs don't think an article is important, they delete it. That's why poets important to the development of a language and culture sometimes don't even have an entry, let alone more that 3 lines while some third-grade rapper that made some noise 10 years back has an essay with 10 000 words covering every detail of their private life.

I've seen flat-out bullshit on wikipedia more than once, I've corrected some things, roughly 30% get rolled back. If an area of expertise has asshole/dimwitt chief editors (or whatever they are called in wikipedia-speak) I often just give up and don't bother.

Wikipedia is a reflection of our times and what's important to us. And it should be viewed as such. With a pound of salt.

... won't be spared. I'm down 20k from my last salary and with AI my productivity has risen 5x. On to of that, the processes I was supposed to automate with code are getting replaced by AI themselves.

Prepare for incoming.

I have never detected even a single advantage of systemd. It didn't bother me enough tow switch distributions, but that's the best I can say for it.

... to my AI metasubscription now. AI does what I ask it to do, I just review the changes and commit. It's like having a personal team of 10-20 experts sitting in a chat just ready to do my bidding. It's not sitting but it doesn't feel like that too often yet.

However it's quite staggering to watch am AI so your job an order of magnitude better than yourself. And that for a bunch of software stacks a human couldn't dream to comprehend. It's also sobering to watch the value-add chains I'm supposed to automate with code being voided entirely by AI. Not only is my job gone, the context with which it makes sense is also rapidly vanishing. You should see the look on the face of the lawyers I work with when the realize how AI does away with them too.

I'm very likely going to leave my current team. I'm in the process of leaving classic Web software development as a day job. ... You guys can't imagine how glad I am not having just software and the Web as my only field of experience and expertise.

Not surprising. Tesla is its 5th generation (AI5) processor, currently manufactured by Samsung and TSMC. I suppose they imagine there are others that will want to use these for their own purposes. Musk is creating his own supply of chips for SpaceX at his TX Terafab. Having terrestrial customers to absorb some of the supply and provide revenue as that ramps up the obvious thing to do.