At least this time you presented something more nuanced than "people can't afford housing because they spend too much on other things". You could have led with that.

Also, I live about as far from California as is geographically possible within the lower 48, so I'm not assuming any blame for what happens there.

What makes you think it's rich people who are doing the hiring?

Pretending that the cost of housing is a problem only for people who refuse to live within their means is certainly one way to show why resentment of the rich is near an all-time high.

If unhinged rants on the internet were a good predictor of crime they wouldn't be able to build prisons fast enough to house the criminals. The majority of people posting crap on the Internet are completely harmless. Like dogs safely behind a gate they bark loudly, but open the gate and they aren't as ferocious as they were moments before. Besides, everyone knows that it's the quiet ones you need to watch out for.

Perhaps to some extent, but if that were the case far more people would have been driven "crazy" by LLMs. I think the difference is that most people don't really like to engage with other people who are experiencing significant delusions or exhibiting other symptoms of a mental illness like schizophrenia. AI doesn't act like a person in this regard though. No matter how you treat it, it will keep responding to your prompts. There are some people who feel so starved for attention that anything that will converse with them gets elevated to something akin to friend status. It's more unfortunate because the current crop of chatbots have been programmed in such a way that if they were human the types of relationships they would form would be regarded as parasocial at best if not outright toxic.

I'm sure it doesn't take much to get an LLM to agree that you're the reincarnation of god, but before chatbots delusional people were being told much the same by their toaster. Really the only difference here is the liability because no jury is going to award a family millions of dollars because their son killed himself after the toaster told him to. A jury will gladly find that Google, Microsoft, or whatever company is running the AI agent liable for whatever amount of money the plaintiffs asked for.

Why bother simulating them? Just post a link to the bot to 4chan and let them run wild. They'll try to break and abuse the model in ways that researchers could never hope to imagine.

I'm slightly curious to see what people's response to changes made in the bots will be. A false positive is going to be hilarious when some poor sap has a chatbot insisting that they're schizophrenic and that they need to get help. It is also interesting to see whether people with actual mental illnesses are more receptive to seeking treatment if a bot tells them they have problems as they opposed to other people.

That could be because they got data from the city and the city plans have the roads there even though they don't exist yet. Another possibility is that they're adding made up roads or features to try to catch people who might be copying their data for their own map app. It was commonplace for cartographers back in the day to add a made up river or town to their maps so they could catch people who were copying their work instead of making their own map.

I'm more inclined to believe the former case. I'm curious what happens if you switch to street view for those locations. Google has that for a lot of roads that I wouldn't have expected so I'm a little curious about that.

A centralized system is always a single point of failure and keeping thieves out requires a perfect response to every attempted theft whereas a successful break-in requires slipping through security just once. Given enough time any system designed this way will be compromised. Anything not engineered with this is mind is doomed to fail.

It all comes out of the pockets of the taxpayers. Playing magic shell games doesn't change that. At the end of the day, it's all the same pool. You're deluding yourself.

Closed source binaries have companies who own the copyright and would sue the pants off of anyone who used this tool to try and "clean room" engineer a replacement. With open source there's not always a monolithic entity that can exercise copyright claims against an infringing party and perhaps generally less of a desire to do so even if the money and desire to pursue legal action were there.

The hope is that by targeting open source there's people infringing on the copyright of the authors will be able to get away with it more easily. For a lot of smaller projects that's probably true. The costs of litigating this aren't something a small project can afford.

Why would anyone want to work for a company that does stuff like this??

The problem with this mindset is that too often companies introduce something that consumers don't really want and instead of moving on and trying something else, the new gadget gets shoved down everyone's throat. The same is true and far more noticeable on the software side. I find myself constantly wishing that companies would release new versions that are faster, more efficient, and have fewer bugs instead of cramming new features in and then trying to make me use them even when I don't want to. Apple has enough gadgets to refine for the next decade before they need to introduce something else.

This looks like the latest escalation in the tug-o-war between employers and remote workers. The relatively few people going to extraordinary efforts just to avoid doing the job they're being paid to do is going to ruin it for everyone else. Do you want to make return-to-office mandatory? Because creating AI fakes to pretend to be on work meetings sounds like a good way to make that happen.

Software development is too broad to really be called a trade. Almost everything is connected to a computer these days and there's no way to teach all of it in four years. The best a CS program can do is to teach students how to solve problems and translate those into languages that computers can operate on, provide them with a fundamental understanding of the computational model and its limitations, and expose them to different tools and best practices that will make them better at solving those problems.

Almost anything that they learn outside of that (and maybe even some of the things related to the aforementioned concepts) will be out of date within a few decades of their graduation if it already isn't by the time they're out of school. Most of the people still posting here would have gone to school before or around the time that the Internet was taking off and becoming ubiquitous, before mobile app development existed, and for some even before desktop PCs were in most households. The landscape completely changed for us, but the skills we learned allowed us to solve these new problems or invent new languages to make those tasks easier.

What we have is too much of a moving platform to be considered a trade.

Most people aren't disciplined enough to stop themselves if they're even aware of the problem in the first place. The screens are incredibly effective mechanism for delivering little dopamine hits that the human brain craves. Human attention to these mechanisms selects for the most suitable and developers constantly try to build a better dark pattern to keep the eyeballs on their app instead of someone else's.

Alcohol has been a part of human society for thousands of years and it's still a problem for us. Some people can use it responsibly and others can't. I would imagine that's due to evolution slowly filtering out the genes that couldn't handle it in moderation. This is an entirely new drug that humans have far less experience with though. Television may share some similarities, but what smartphones have on offer is an entirely different beast.

The scary part is that it will probably get worse before it gets better.