I think there are plenty of people who want to help, in principle. People also tend to get overwhelmed in a crisis situation and suffer extreme performance anxiety. It's not like we don't know that there are biological underpinnings for this. Consider the effects of epinephrine/adrenaline. It boosts some senses, making you more alert to danger while dulling sensations of physical pain, etc., priming you for fight or flight. However, it also compromises your higher cognitive abilities, memory, etc. PET scans have demonstrated drops in blood flow to areas of the brain handling those things. That is why people trained for emergency situations (EMTs, firefighters, soldiers, etc.) often drill and drill and drill and do so when possible in situations that are close to real emergencies as possible. Such training may reduce the level of epinephrine produced in such situations, or at least increase tolerance to it, while also converting the behaviors required in those situations from ones that require a lot of cognition to ones that are as close to natural instincts as possible. The simple fact is that most people don't have that kind of training and the average person, not very smart to start with, drops a hefty chunk of IQ points in a crisis.

That is only one reason people are bad in a crisis. I already mentioned the group dynamics issue where people in a group can often end up waiting for someone else to act so they can follow, lack confidence in their own abilities, etc., etc. There is also the concern about liability, etc. and an AC that replied to you also said pretty much exactly that in their post: that they had been burned before trying to help and now refuse to.

Right, but I think that would be an example of what I am talking about really. You were upset that their attention was constantly focused on you and it felt aggressive and unsettling. I think it is likely it would have been the same if they were simply staring directly at you, especially right at your eyes.

Well, that's the thing isn't it. The answer to that question is basically everyone. Restaurant, theater, retail store, bank, DMV, anywhere along the road in any public area, etc.; if you go out to eat, you are on camera basically the whole time. Virtually every business has a camera on you, including restaurants. You could argue that it's not pointed directly at you, but that's meaningless for a security camera. If you're in frame, it's pointed at you. The human eye is different. It has high resolution concentrated in central vision, with low resolution in peripheral vision. Being pointed directly at you means something and we instinctively know if means something because of millions and millions of years of evolution that have programmed us to instinctively understand what eye focus can mean. Now, the ubiquitous cameras also provoke those same instincts, but at a lower level. Admittedly one that is causing many people some degree of constant stress, but a lot of people have simply accepted that (and its continuous expansion) as the status quo.

Sure, it's creepy. It is pretty necessary for it to be legal though. There are a lot of very good reasons to allow recording people in public. The fact that it is legal is the only reason that, for example, police actions can be recorded in public. Even when it is legal, we still hear plenty of stories about police officers arresting people, assaulting them, stealing their property and destroying it, etc., etc. recording. Ultimately, the social good is largely inseparable from the social bad. Also, much of the behavior that people object to with cameras recording them, they would probably object to equally as much if it were done without a camera. For example someone on a train being stared at intently by a stranger is probably also going to feel like the person is creepy even when the person is not using a camera. Ultimately, if that kind of behavior goes too far, with or without a camera, there are legal concepts like harassment that can apply. It's not all clean an cut and dried, but I would say that, in a lot of cases, it is not really a camera recording that is the primary problem.

Hmm. I've been watching TV for most of my life. Based on thousands upon thousands of examples of people filming in public places, that absolutely, 100% happens a lot. There may be nuances like news programs trying to get releases for people they interview directly or focus on, but that's often after the fact and not really always done.

I'm not making that assumption though. Clearly people sometimes get upset about that also. I am just noting that it seems like the vitriol over these glasses seems to be a bit higher than over other recording devices, or at least it seems like people are classing them differently. Consider the fact that, currently, there are five replies and one flamebait moderation on my fairly modest post. I mean, reading my post, does it seem like flamebait? Nevertheless, it appears that the mere suggestion that people's response to these devices might be a little elevated made someone angry enough to waste a mod point on that. I have seen plenty of people suggest that such glasses type devices should be outlawed while not demanding that cell phone cameras, etc. be outlawed. There can be other reasons for that of course. Other forms of recording devices are considered to be relatively established at this point, whereas devices like this are relatively "new" (not really new at this point). I still think that the primal aggression of direct gaze may be a factor, but I am not saying it is the only factor.

It's sad to say, but this is part and parcel for the US. The rich want to get every advantage they can, and their money allows them to. So as students with real disabilities get accommodations the rich see it as someone getting something they don't and immediately go about finding a way they can get it to. It doesn't matter that they don't deserve it, they think they deserve it simply because someone else is getting it. And with the US healthcare system, they can always find someone willing to give their kids a diagnosis whether they need it or not because they are willing to pay the price (And can afford to pay the price). This is what the US has become, a plutocracy. The rich get whatever they want because they can afford to buy everything and in the US, everything is for sale.

There are still places I write out checks because I get a discount for doing it.

Given this level of subterfuge, all I can say is, "Sloppy job, Mossad."

Almost as bad as the people who try to get their iPhone payment app running.

There is a serious question of why people get so upset about glasses that record as opposed to cell phones, etc. I have a theory on it and that is that, despite all the emphasis on eye contact, humans actually find it really aggressive and threatening. Traditional camcorders, cameras, and taking cell phone video all either outright block the eyes or at least the eyeline. You are not staring at them, you are staring at your phone. I think that is, at least in part why devices like smart glasses make people more aggressive than other recording.

The clear counterargument to that though, in the case of face mounted recording devices, is that they don't force a choice between helping/calling police or filming. True, that should not be a real dilemma, the obvious choice should be to help. The thing is, that problem in crowds of no-one stepping forward to help does not exist purely because of people recording. There are lots of reasons people don't put themselves forward in situations like that. One of them is the assumption that someone else will be able to handle it better than them, or simply waiting for someone to step forward and lead, etc. There are also concerns about liability, possibly about self-endangerment and a dozen or more anxieties and neuroses that can cause the problem. Recording with a cell phone probably exacerbates the issue though, by giving people an activity that, in the moment, their mind can rationalize as doing something. If recording is a less active and more passive process, leaving people free to actually do something and taking away their excuse for rationalization, it might encourage more people to actually help.

Mileage may vary, of course. On balance though, it seems like head mounted recording devices are more of a solution to the problem you were talking about than an exemplar of it.

First,mandatory screen time needs to be limited. If they want text books in ebook form, great, but they'll need a way to restrict school issued pads to school work during the school day.

On the flip side, I have more than once heard a parent complaining that homework is being given that requires a computer to complete where a school doesn't allow chromebooks to be taken home. That's equally absurd. Not every family can afford to give each kid a computer, and sometimes computers break. It's not like parents can just grab an extra one at the corner store like they would a pack of pencils or paper. If school work requires a computer and/or internet connection, the school should provide it. If that includes homework, the students must be allowed to take it home.

If the schools don't like that or can't afford it, they can issue text books and homework that can be completed with pencil and paper (yes, that includes accepting hand written essays).

And as for not letting parents view the assignments, that's ridiculous. Of course the parents have a right to see it. If some company wants to claim that to be proprietary information, I guess the school can't use it at all.

It's crazy to complain about students on their screens too much and then have mandatory screen time. It's equally ridiculous to complain that parents need to be more involved and then shut parents out.

/rant

This is "absolutely without question" incorrect. One of the most useful properties of LLMs is demonstrated in-context learning capabilities where a good instruction tuned model is able to learn from conversations and information provided to it without modifying model weights.

Who cares? This is a rather specific and strange distinction without a difference that does not seem to be in any way related to anything stated in this thread. Randomness in token selection impacts the KV matrix which impacts evaluation of subsequent tokens.

This is pure BS, key value matrices are maintained throughout.

Again with determinism nonsense.

LOL see the program isn't malfunctioning it is just doing what it was programmed to do. These word games are pointless.

The dot.com bubble provided value in the form of useful infrastructure investments. When the AI bubble bursts all you are going to be left with are rooms full of densely packed GPUs that will be scrapped and sold off for pennies on the dollar.

How much of a time saver is it to have a magical oracle at your fingertips that constantly lies to you? How much time is saved when you have to externally cross check everything it says? It only saves "tremendous" time when you can afford not to care about the results.

All of these post training bludgeons are inherently dishonest. They attempt to sell an intentional lie inserted by model designers and ultimately make the technology worse and more difficult to use.