You've got some weird fantasies/fetishes, dude.

Also solid advice, which doesn't require bulk public surveilance.

Oh piss off with your "think of the children" nonsense. If you're so worried about your daughter there are countless ways to help keep her safe that don't involve the government warrantlessly tracking the entire population. Teaching her how to properly knee someone in the groin is far more effective anyway.

Spot on. I share the same assumptions as you, that it's not that hard for someone who does such things for a living to figure out how to get the model to recognize "hey, there's water there" and "hey, there's a shitload of water there, probably shouldn't barrel through it".

My guess? I doubt it saw or recognized the intent of the hand gesture, but it almost certainly recognized the flashing red. I assume the "thought" process was "well, nobody else is going. We all stopped at roughly the same time. Yeehaw." but who knows. Doesn't Tesla have some sort of "playback" feature where it can show you what it saw? Or is that only a real-time view?

Fun, spectacularly interesting, and terrifying, all at the same time.

This, right here. Fun story from last week: I had a guy in front of me heading up an onramp. He literally drove OFF the ramp, three wheels in the grass, before he jerked back onto the ramp. Once he got on the highway, did the same thing twice. 80% into another lane before violently jerking back into the lane he was supposed to be in. He was turned completely around in the driver's seat digging for something in the back. I went around the dickhead, gave him a little "howbout you drive?" honk on the way by. The dude literally followed me for a mile with his middle finger hanging out the window. Then promptly went back to digging for whatever was more important than the safety of him and his fellow drivers.

Do autonomous "drivers" fuck up? Of course they do. But they can be "taught" to not do it again. And they certainly aren't turned completely around in their drivers seat, digging for something in the back of their car, while they are driving down the freeway. I'll take the AI driver, thank you very much.

My assumption is that the geo-fencing is a short-term "implement it immediately" kind of thing, while a "teach the Borg to not drive into the water" is a much longer lead time. If the sensor suite is capable of detecting water (which I have no idea what sensors they even have on them, nor their capabilities) I assume it's a relatively easy fix. Just like every other "don't do X" fix that's been implemented after something bad happens. If the sensors can't see water, that's a way different problem.

We don't directly pay licensing for OTA here. It's mainly supported by "OTA Fees" paid by cable/satellite subscribers. (I think, anyway. I know for sure there's a fee for cable/satellite, but no clue on the "mainly" part of my comment.)

Anyway, the prime reason for my OTA setup is for Football season (the American kind). I spent a couple hundred bucks once to set it up, a dual Silicon Dust IP tuner, good antenna, and plex-pass on sale. Now I can watch the games with no additional cost.

No, they just don't have any other words available in their vocabulary. Check their comment history, it reads like a hormonal teenager's Twitter feed.

No. This isn't "sloppy text editing", this is submitting unverified information. Including copy-pasted generative AI content that wasn't verified prior to submission is, by very definition, submitting unverified information. They are saying that the penalty for submitting unverified AI content is as severe as the penalty for submitting any other unverified information. The entire premise is that if you are willing to submit unverified generative AI content, what else is in the paper that wasn't verified?

No. It's like getting busted submitting garbage to an organization whose entire business model revolves around accepting quality work, then not being allowed to submit your work to that organization because you are known to have submitted garbage. Or getting banned from a sports league after you get caught cheating. Or having your drivers license pulled when you get busted for drunk driving. Or getting fired for not doing your job. There's no "slippery slope" here. This is strictly a "your actions have consequences" thing.

My entire comment was replying to the "How does an AI experience external stimuli?" portion of GP's comment. That's it. And that's literally what an LLM does. I provide it stimuli, in the form of text, voice, picture, etc. and it responds. The basic definition of "external simuli".

You're "gain an understanding of physics from video" is adding criteria beyond the scope of the conversation. Whether or not an LLM currently can, or will ever be able to, interpret the content of a video is beyond what I'm willing to get into. Hell, we could probably spend the next 3 days just discussing what you mean by "gain an understanding of physics", and that's just a line of thought I'm not willing to expend the braincells on.

By what you type into it.

Sure you can. I literally just tried it. I have a note on my desk for a book I need to order. The note says: "Half Price Books" "ISBN" "xxxxxxxxxxxxx". It's on a "joke" sticky note that has a little picture of a toaster in a bath tub and says "Live Laugh Toaster Bath" at the bottom. I took a picture of it and sent it to Gemini: "The handwritten note provides the details for a specific book." Then lists the title, ISBN, Author, and the store I referenced. Then goes on to say "The note also features a humorous, satirical "Live Laugh Toaster Bath" cartoon graphic printed along the side margin"

Of course it can't actually "see" anything, as it doesn't have eyes, but it is perfectly capable of processing information in image form.

You're being pedantic. Of course it isn't a "sense" in the way you and I can touch/feel/taste/see, but the words (or image) "monkey-patched" into it are absolutely external stimuli.

Not necessarily. Sorry for the AI copy, but it's a better summary than I can write: "A typical concrete specification, such as ASTM C94, generally limits a truck mixer to a maximum of 300 total revolutions from the time water is added to the cement and aggregates until complete discharge." There's also a 90-minute limit that starts as soon as you add water.

I've participated in a handful of huge pours where they batched offsite. (I don't recall how many cubic yards, order of magnitude around 7,000) But this was pretty rural, so traffic wasn't generally a concern. I want to say they batched about 30 minutes away. If the truck is getting near the 90-minute mark they can "kill" the batch, slowing/stopping the hydration. It renders the concrete useless, but at least saves the truck. I didn't realize this, but apparently about 20-40 lbs of plain white sugar will accomplish the task.

Trust me, I'm not holding my breath. The insurance companies here make too much money to just let that cash cow wander out to pasture.

As was OPs.

This is a good thing. It would be awesome if coal wasn't growing, but "growing more slowly" is good thing.

Yup. So far. The world still needs power, that isn't going to change. All we can do is continue to push for sources that are more conducive to our continued habitation of this planet.

Agreed. It's an abomination. Single Payer is the only solution.