Rhaedas

We’ll find a remote island somewhere and name it Karma.

I’m going to take this from a different angle. These companies have over the years scraped everything they could get their hands on to build their models, and given the volume, most of that is unlikely to have been vetted well, if at all. So they’ve been poisoning the LLMs themselves in the rush to get the best thing out there before others do, and that’s why we get the shit we get in the middle of some amazing achievements. The very fact that they’ve been growing these models not with cultivation principles but with guardrails says everything about the core source’s tainted condition.

Right. I actually don’t have an issue with them not having a transponder on. That was the least of the problems. But not being aware of standard traffic around an airport… come on.

That’s comforting that they could be that way and be running silent from ATC and other aircraft anywhere. What could go wrong?

Nothing will happen because it’s US military, but shame on those pilots because you know THEY saw the JetBlue coming their way, knew the route they’d take, and didn’t give a shit. They didn’t even have to say anything, just change course.

If it’s slow, then it’s the central backup and you use anything else for regular use. Just having it as a fallback for recovery would be huge.

Somehow I managed to get inspired and created/published a few medium content books (aka puzzle and other oddities in niche areas) online. Actually sold a couple somehow, but if they don’t do well that’s okay. Trying to work on a real novel now. Not really a writer at heart, but have fallen in love with an idea and the deeper I get the more I want it to succeed. Ironically I took a break from the writing part and played around with cover ideas and character envisioning. Was better at that since I’m a graphics person, and finishing that side trip inspired me to jump back into the writing.

You do some beautiful work. As a kid I collected a few of the metal miniatures from the board game, loved them. Models don’t have to be big to capture the imagination.

And they were right. That’s why it’s called universal basic income, everyone gets the same amount (well, probably adjusted based on cost of living, but ideally the same). What they’re implying is not correct, that everyone will get an amount that they’ll be happy with. Those that work and get UBI will be both comfortable and secure if things go wrong. Those who can’t or don’t want to work (maybe for a period of time, maybe they’re just done with working) will have enough to “basically” survive. No one is getting rich off UBI, but everyone will be better off that isn’t now.

The real important thing that UBI can bring is making companies have to do more to convince people to work for them. Get rid of health insurance tied to the workplace (like with single payer) and then they have to offer real benefits and good working conditions. They’ll fight it screaming and kicking too, they like how workers have to play their game and take what they’re willing to give, not the other way around.

Sorry for the long diatribe. I guess I have strong feelings about it.

I rarely downvote unless it’s an obvious troll. I’ve always seen downvotes as meaning not related to the topic, not a negative opinion. If it’s worth engaging I’ll introduce my own take and why I think they’re wrong. For discussion purposes, as that’s why I’ve always been on such places starting back when they were called discussion boards. I will upvote for something I agree with, but for the purpose of it hopefully seeing more light. I’ll comment on it as well, trying to avoid just a “me too”, although sometimes I think my longer comment may as well be that sometimes. I try to bring something new if I can.

My biggest problem is that I engage so much in the topic and comments I often forget to upvote the main post. I hope anything I do within has the same effect. In theory more comments helps, right?

That has its own slope of discrimination from data due to being able to pay or not. If we determine a certain thing is okay ethically to screen for, anyone should be able to get it. Bad enough to have one gray area, we don’t need a gradient of gray everywhere.

My first question was, why is this a target? This is terrorism and outside war rules (insane that we have rules for war instead of just not having it). Laws are only as good as their enforcement though.

Only thing I can think of while sitting comfortably at the keyboard and not in that car panicking is to kick out the windshield and hope you can swim under and out after the water rushes in. If the car is resting at the bottom and that’s blocked… no, I can’t think of one. I don’t imagine cars are very airtight, especially since they drowned, so there must not have been an air pocket for them to last a few hours.

You’re correct on their limitations. That doesn’t stop corporations from implementing them, sometimes as an extra tool, sometimes as a rash displacement of paid labor, and often without your last step, checking the results they output.

LLMs are a specialized tool, but CEOs are using it as a hammer where they see nails everywhere, and it has displaced some workers. A few have realized the mistake and backtracked, but they didn’t necessarily put workers back. As per usual anytime there is displacement.

And for the record, while LLMs are technically under the general AI classification, they are not AI in the sense of what the term AI brings to the mind (AGI). But they have definitely been marketed as such because what started as AI research turned into a money grab that is still going on.

Useful maybe. For what purposes though… getting labor costs down, pumping out stuff fast assuming it’s correct because it’s AI, being ahead of their competitors. Useful as in productive? Maybe for some cases when they know what AI can and can’t do or its limitations. I get the impression from this year’s news stories that a lot of them jumped on it because it was the new thing, following everyone else. A lot got burned, some backtracked where they could, some are quiet but aren’t pursuing it as much as they advertised.

OP is right, companies will go the direction they feel consumers will buy more from, and if that’s a “No AI” slogan, that’s what they’ll put. There’s no regulations on it, so just like before with ingredients or other labeling before rules were set, they’ll lie to get you to buy it. Hell, from a software pov there’s a big thing now on apps being sold as “FOSS” that are not, because there’s no rules to govern it. Caveat emptor.

Sometimes? I wouldn’t worry about your unease in walking out of your bank with your money, I’d worry about having your money in that back where they apparently give money out to anyone.

Seems a bit too easy if that’s all you’re doing without any verification. If they’re checking that you are who you say you are, then it’s your money.

I was in the webmaster role for a website from the early start of the internet - SEO started off as simple ways to help improve index placement by giving different methods to the web creators to aid in better categorization of content. It quickly became an arms race of how to best game the system, and the system kept changing as well because the old SEO basics like keyword and content arrangement wasn’t enough. There was one search engine I participated in (I can’t recall now which one) that did the pay for clicks, and you’d literally have to pump money in the online app to try and stay above your keyword competitors, all in real time. It got stupid. And I got frustrated with it, as I felt the original goal to find the best website for a particular search had been long lost and now it was all about mechanisms to profit from everyone trying to make that first page hit. The “best” sites that couldn’t play this game were lost.

Google became the dominant player by buying up other databases and engines, but even with this gaming they used to be able to produce results if you knew how to phrase searches beyond just a few words. It’s almost like the whole AI prompting, what you put in makes a difference. But they eventually changed things and started getting worse results, lots of duplication, and then added AI which ruined anything they still had of quality.

I miss Hotbot. That was my go-to long ago, and it was so good. It became part of Google eventually.

I have four, all in mid-grade school (7-8).

A mobile of various paper models of satellites, along with a research paper that told about them.

A cardboard model of the USS Monitor from the Civil War (for US History obviously).

Another for history was a functioning balsa wood model of a guillotine, with a (dull) metal blade. And a deheaded G.I. Joe (I didn’t have any French aristocrat dolls handy).

A video book report made by with a few friends using the library’s video camera (back before phone cameras). We did it in the style of a satirical news program/Monty Python humor with various clips from reporters of parts of the book’s story. I don’t know why I never asked for a copy… but you don’t think about that as a kid.

And that gentlemen, is why we don’t use test and push right out to production. The consumer is far better at finding flaws than we are.