No, that's like saying that it's inevitable that indoor plumbing and air conditioning will continue to spread and be adopted by everyone who can afford them. Or that the police will use national computer databases to track criminals, and helicopters for urban surveillance and pursuit. Or that the military is going to use more drones in the future.
I believe the collection of the information is inevitable. What I would push for instead of driving them to make the cameras and databases more clandestine than they already are is for the information that they collect to be made openly available to all.
As things are, it's a very asymmetrical power tool for the advantage of the (government) operators.
When ALL the information is available to everyone, we can talk about where the cameras do and do not need to be. And any unapproved cameras can be suppressed as evidence against private individuals.
Can you say: "conflict of interest"? We're at trial, the cop(s) who performed the arrest made a judgement call in the field - of course they're going to double down. What would it do for the career of a cop on the stand to say "you know, I think we made a mistake that day..."? The fact that the case has gone to trial basically makes the cop's testimony redundant, what they're going to say is basically a foregone conclusion, why waste time making them say it again?
For what memory chips cost these days, they should record continuously anytime the camera (accelerometer in the camera) detects motion within the previous 10 minutes. If they're on-body, or in a moving car, they should be recording.
The "save" button could work the same: mark 30 seconds before until "save" is deactivated to be "do not delete this for rotation" - but otherwise, save everything anyway, only rotate out after 2TB of memory card is full, and download at the end of every shift.
Better still, download continuously to the car and 5G it to a cloud server where the department can't delete it.
I took a speeding ticket to court, had the officer sitting behind me pre-trial talkin' smack with a colleague "why are you here? Speeding, ha, how hard is that?" Yeah, so he gets on the stand and "reads from his notes" every single thing he said was fabricated, only my location was accurate, his location was a lie: in reality he "witnessed" me from a side street 3 blocks back from the intersection he crossed but in his testimony he "observed me passing a line of five cars" - yeah, except that never happened, what I was passing was a single gardening truck doing 10mph for the past 3 blocks, the other 4 cars were stacked up behind me.
I'd be in favor of a "private" button that they can press for such circumstances. The video is still recorded, but marked private - plays back black and silent on ordinary playback software. If it's ever in legitimate question of whether or not "private" was pressed inappropriately the private video can be restored to full picture and sound with the appropriate code key.
The technical solution is: body cam footage is automatically, frequently, uploaded to cloud servers that the department does not control. The department gets read-only access, nobody gets the ability to delete footage for 7 years, and defense attorneys get automatic access to everything remotely related to their case.
Also: planting evidence and sending the falsely accused to prison for 6 months is a misdemeanor punished with suspended sentences and probation? That department owes the falsely accused damages for lost wages and damage to their ability to obtain future employment. That's actually a "superpower" cops know all too well: if you've never been arrested they can seriously screw up your life with absolute impunity just by arresting you - charges never have to be filed, that arrest on your record - however baseless it may be - can hurt you in all sorts of ways, especially employability, for the rest of your life.
ML already has demonstrated tremendous capability increases for automated machines, starting with postal letter sorters decades ago, proceeding through ever more advanced (and still limited, occasionally flawed - like people) image recognition.
LLM puts more of a "natural language interface" on things, making phone trees into something less infuriating to use and ultimately more helpful.
LLMs, which are too costly to train and run
That's a matter of application
inherently too unreliable for safety-critical or health-critical use
Yeah, although I can see LLMs being helpful as a front end, in addition to the traditional checklist systems used for safety regulation, medical Dx and other guidance, an LLM can (and has, for me) provided (incomplete, sometimes flawed) targeted insights into material it reviews - improving the human review process as an adjunct tool, not as a replacement for the human reviewer.
too flaky for any use requiring auditability
Definitely. Mostly I have been using LLM generated code to create deterministic processes which can be verified as correct - it's pretty good at that, I could write the same code myself but the AI agent/LLM can write that kind of (simple) program 5x-10x faster for 10% of the "brain fatigue" and I can focus on the real problems we're trying to solve. Having those deterministic tools again makes review and evaluation of large spreadsheets a more thorough and less labor intense process. People make mistakes, too, and when you give them (for this morning's example) a spreadsheet with 2000 rows and 30 columns to "evaluate" - beyond people's "context window capacity" as well... we need tools that focus on the important 50 lines and 8 columns without missing the occasional rare important datapoints...
So far, with LLMs, the game ain’t worth the candle,
The better modern models, in roughly the past 10 months or so, have turned a corner for some computer programming tasks, and over those 10 months they have improved rather significantly. It's not the panacea revolution that a lot of breathless journalists describe, but it's a better tool assisting in the creation of simple programs (and simple components of larger programs) than anything I have used in the previous 45 years, and over the past 10 months the level of complexity / size of programs the LLMs can effectively handle has roughly tripled, in my estimation for my applications.
even without considering the enormous environmental damage caused by their supporting infrastructure.
When it's used for worthless garbage (as most of it seems to be today), I agree with this evaluation. Focused on good use cases? In specifically good use cases, the power / environmental impacts range from trivial to positive - in those cases where the AI agents/LLMs are saving human labor - human labor and its infrastructure has enormous environmental impact too.
And this is my point: the (super) human and the machine are both capable of infringing copyright - breaking the law. The question is: are they actually doing it?
If you sit the human down with a researcher and they write out: HP5 Goblet of Fire in its entirety 99.9%+ accurately (for the edition they are recalling) - that's research, fair use. As was done with the AI models by some researchers. Are the AI models out there in the real world also selling copies of their training books in full, or substantial parts, to their users? I haven't seen demonstration of that, yet.
Curious: Does Montreal allow "employment at will" like most states in the US do? If it does, I can't imagine an UbiSoft contract not including it. The article definitely makes it sound like a termination for cause, but what's written on his termination paperwork may be entirely different.
Not just that, but "working with your hands" has been seen all kinds of machines automating people out of jobs for the past 200+ years, AI/LLM will only make automation more capable, and more undercutting of people's manual labor costs.
True, but in this case it seems worth doing due to the relatively patient, selective nature of the attack - it would at least clean out a compromised Notepad++ if it had not spread to a wider system compromise yet.
Point is: some humans can do this without a machine. If a human is assisted by a machine to do something that other humans can do but they cannot - that is illegal?
That's what I read in the article - the "researchers" may have had other interfaces they were using. Also, since that "research" came out, I suspect the models have compensated to prevent the appearance of copying...
Just give me the prompt and model that will generate an entire Harry Potter book so I can check it out.
Start with the first line of the book (enough that it won't be confused with other material in the training set...) the LLM will return some of the next line. Feed it that and it will return some of what comes next, rinse, lather, repeat - researchers have gotten significant chunks of novels regurgitated this way.
You may not have photographic memory, but dozens of flesh and blood humans do. Are they "illegal" to exist? They can read a book then recite it back to you.
No, that's like saying that it's inevitable that indoor plumbing and air conditioning will continue to spread and be adopted by everyone who can afford them. Or that the police will use national computer databases to track criminals, and helicopters for urban surveillance and pursuit. Or that the military is going to use more drones in the future.
Even before everyone carried GPS trackers in their pockets and digital cameras became dirt cheap, you were being tracked and analyzed by your credit card: https://www.forbes.com/sites/kashmirhill/2012/02/16/how-target-figured-out-a-teen-girl-was-pregnant-before-her-father-did/