To not require expensive re-training of models for just the purpose of yet-another-slightly-different CPU or micro-controller becoming a relevant target.

I currently work in a "Fortune 500" company, and people left and right are trusting LLM-bots with both writing code and reviewing the resulting PRs, for things one could consider "important tasks". And yes, this utterly irresponsible, but cheap and convenient, therefore it is done. People who don't want to participate in the AI-Slop-production will leave the company. I know I will.

Then take a look how many people privately trust "AI agents" with everything from their emails to their banking. Madness, yes, but there is no denying this is happening.

I happen to know someone who is working on air-traffic-control IT systems. And his boss already found a loop-hole to make the AI Slop happen, there, too: Just more outsourcing to external companies, and then some internal employee, who of course has no reasonable chance to actually properly review the externally supplied code, is asked to "sign off" on their contributions.

Sony sold "Qualia" TVs with RGB LED backlight in 2004: https://www.sony.com/en/SonyIn...

It is ridiculous this is now touted as a new technology. (And no, I am not ever buying Sony again, I don't want their root-kit to compromise my hardware.)

With the exception of machine language, all contemporary programming languages where designed for humans to use. Now that the era of humans being able and willing to write source code is ending(*), and more and more people relying on LLM-based bots to write software instead, it is easy to predict that the most used "programming languages" of the future will be one tailored not to be used by humans, but by bots. Like this "programming language for transformers", for example.

(*) Disclaimer: I am not saying this a desirable future, but the signs are everywhere that it is coming.

And the movie was a pretty bland Western without any interesting ideas... as entertaining as watching Iron oxide.

... when it was new - guess I am now a qualified Air Traffic Controller, and we can skip the training :-)

For those who also want to qualify: Kennedy Approach.

I agree, and would also be much more convinced of the theory if the Schwinger effect had finally been demonstrated experimentally, rather than having yet-another Breit-Wheeler process demonstration.

Remember the Evil Bit from RFC 3514, and how it got us rid of all malicious network traffic, once and for all?

Irresponsible people (including irresponsible employees) have evidently shown a lot of interest in automating away all kinds of stuff by installing "OpenClaw" and giving it access to all kinds of sensitive information and credentials to act on their behalf. If you run "OpenClaw" in some sandbox that does not give it access to all that sensitive information and credentials, the purpose of "get it done for me, I cannot be bothered" will not get fulfilled, and therefore that sandbox will either be intentionally be circumvented or the bot becomes useless in comparison to what people (ab)use OpenClaw for. No kind of "security layer" can change that.

... only to learn later that the thing you paid for was largely made in Asia, with only some packaging or branding happening in the US? It seems kind of ironic and well-deserved that the same kind of deceit now also happens to companies that want to buy "made in USA" labor.

Power electronics have become cheap enough that inverters, including ones that synchronize to an existing network, are quite easy to manufacture even for home use. And yes, running multiple such in parallel is possible, and actually supported by some of the "balcony PV"-inverters sold in Germany, not least to sneakily circumvent the 800W regulatory limit. You can even mix balcony PV with multiple home battery units, if you want. Of course one still needs to take care that the cabling won't be overloaded, and especially the battery units are still so expensive that they need to last way longer than their warranty time before you can break even.

No, that would be true only if all balcony panels were battery-buffered in one's home. But buffer batteries for home use are still quite expensive, so depending on how many years they actually last, they may or may not break even against the savings from power you did not have to buy. For most people, balcony PV is less of a profit center, and more of a virtue signal "look, I'm doing my part for non-fossil energy!". The utilities of course hate balcony PV, and while the safety issues are just a pretense, they do have a point that balcony PV provides the grid with even more power exactly at the times when there is already more than enough available from the existing PV farms. It would really help if both centralized and decentralized battery storage became cheap enough that every kWh from PV could easily be stored somewhere.

Yes, and in addition to the usual platform lock-in, there will just be way fewer humans capable and willing to do thinking on their own. Applicants will roll their eyes when asked to perform intellectual work instead of delegating it to some LLM. Brains atrophy at least as quick as muscles when not used.

... when the media every so often told us the horror-stories from behind the Iron Curtain, where the evil Stasi or the KGB spied on their own people, and did so as lessons about the difference between "free" and authoritarian countries. But now it seems the "cold war" was thoroughly won by the Warsaw pact states, at least with regards to spreading that habit of excessive eavesdropping on everyone all around the globe.

Given how quickly all sorts of nonsensical web-sites can be created by vibe-coders that don't really know much about AI other than that it is a current hype, it is no wonder so many such sites are competing for attention... and if only a little successful in attracting media coverage, will be swamped by the usual scammers and spammers. "Moltbook" hat a fun-to-talk-about idea, but otherwise is just pure garbage, just like "rent-a-human.ai" and many others.