The super intelligent AI wouldn't ask you to do anything. It would find the right people and trick them into doing the necessary steps, whatever they are.
This is groundless speculation; if you're going to go that far off the map, then we don't need to worry because a good AI (blue eyes not red) will have learnt to love by then and will save us.
There is currently no reason to believe that such a superintelligence is likely or would have any of the powers people claim.
There's no logic here already; the principle you should be applying is Hitchen's razor[1].
There is no reason to assume that super intelligent evil AI with magic powers will appear; while we are purely in the realm of (hackneyed) fantasy, why stop at imagining just one thing? You can build the whole story, not just the basilisk.
The parent comment is a story, not a prediction; it should be treated like one.
>This is groundless speculation; if you're going to go that far off the map, then we don't need to worry because a good AI (blue eyes not red) will have learnt to love by then and will save us.
By shutting down open-weight models? Why not just do it now then?
To me this is strange. In the UK I'm so used to ANPR I just assumed my car is tracked wherever I go. There were a few incidents when the automatic speed cameras first came out, filling them with expanding foam so they exploded etc but even those died out after a while.
Police in the UK are held to a much higher standard, and the culture in the UK is much more tolerant of government control.
I’m also going to guess that there are significant access controls on the systems in the UK that would prevent things like tracking an ex, or cops using ALPR to see if you have visited a healthcare facility that offers abortions among its other services (both real examples).
I could be wrong, but I don't think the UK has a network of privately owned ANPRs that can create ~complete pictures of people's comings and goings, and that data has effectively zero legal protections whatsoever.
Automated speed cameras are fairly common in the US but are obviously not a surveillance network.
I'm not sure I've ever noticed a single one or had a ticket from any or heard of anyone ever getting a ticket from one. They don't seem common in my experience, or maybe it's just like an east coast thing because I've never lived in that part of the US.
Well for one, there's not nearly enough of them to be valuable the way that Flock is.
Two, they're pointed exclusively at roads.
Three, their outputs don't show up in court cases as evidence, which itself dramatically reduces the danger (even if they potentially could maybe support parallel construction)
Speed and stop cameras are very very good and should be virtually everywhere with extremely strict control on their outputs.
> Three, their outputs don't show up in court cases as evidence
That's not true - I was recently a juror on a UK trial where we were shown ANPR photography of the defendants' vehicles. Here's some other examples of cases in the UK where the prosecution presented the jury with ANPR evidence:
You misunderstood the thread. Automated speed cameras are not part of a surveillance network.
ANPR networks are obviously created specifically for surveillance.
GP said "I remember pushback to automated speed cameras" and I said sure, but that was still very different because they weren't part of a surveillance network. Your proof that ANPRs are used for surveillance illustrates the same difference I am pointing out.
Then the gap between UK's government-owned ANPRs and the privately owned Flock cameras is still probably quite large. I don't know the UK's privacy laws, but they probably exist. Flock, as a private party in the US, essentially has zero controls whatsoever on how the data is used. Including – most critically – that the government can simply buy the data from Flock (sans warrant) that would be illegal for the government to produce itself.
You see that I haven't heard of the thing that I explicitly mention in the exact comment you just replied to?
In any case, an evidence modality that can only be used via parallel construction is still far, far, far less risky than one that can be freely introduced in court.
Flock doesn't just sell ALPRs. They bundle it with CCTV, audio sensors, sells to private parties, analyze visual data. That's why they're everywhere. Pointed at pools, walking paths, indoors. In countries where ALPRs are widespread, they're exclusively on roads, because.. they read license plates.
well hopefully this snowballs and most public facing cameras other than private ones by people and businesses go extinct (state run cameras). We don't need to be recording every citizen's movement "just in case" . Society got along for thousands of years without it, and the potential for ever increasing tendency toward authoritarian regimes points to such surveillance as a terrible mistake in any society striving for a democracy.
Just because the UK has become a dystopian hellhole with cameras everywhere, Internet censorship, and repeated attempts are ending anonymity online, does not mean that other places should follow suit without resistance.
Is anyone feeding models touch data? It seems the main thing we want the robots to do is touch things, but we are just feeding them audio/video/images. The model has to learn how to touch things despite never having touched anything before. Perhaps that's why they all look so hesitant when they touch things?
I get that while driving home. I've driven half an hour and then realise I've been thinking about something else the entire time and cannot recall anything about the journey.
I have heard "the poor type safety" argument from writers of strongly typed languages for many many years. Having written js and python for a large amount of my carrier I can count on one hand the number of times I've found a bug that was due to a type issue. With LLMs it has been the same pattern. They don't seem to produce issues with types.
well i’ve come across it loads, especially 1/2) during REPL style build outs and 2/2) calling libraries and frameworks you are not yet familiar with.
perl another offender… is it a hash? is it an arrayref? over time you get it right, but by trial and error and looping. json suffers this too, arrays different from strings, different from numbers etc, but opaque until checked and liable to change
The feeling of being an outsider because you don't know the acronym is just as real in non-public settings. The new guy on the team for example. There's a real feeling that you should already know it, even if it's the first time you've heard the other person say it. The first time you heard GBIF, did you know what it meant? How did you feel when you heard it?
Honestly? I looked it up on the glossary page of the company wiki and didn't feel much of anything about having to learn it.
At the time I didn't know what "Global Biodiversity Information Facility" referred to either, or why it was relevant to my team. Even without the acronym, I still would have had to go look it up or ask about it if someone used it with the expectation I was already familiar with it. I don't think the acronym had any significant effect on my experience of learning the concept.
There's really a feeling now pushing back against learning in general. The feeling is that it is pointless since technology would just do it for you. When I started learning Chinese a friend just wouldn't stop talking about how the latest airpods will just be able to translate for you. It was really rather demoralising. But there is still something incredibly rewarding about having that knowledge in your own head and not having to go find someone or something to ask. I push on regardless.
Funny that your example is Chinese. Have you seen the reports that Chinese universities are cutting back on foreign-language programmes because the expectation is that AI translation is the way of the future?
So, your friend was just expressing the general zeitgeist. But something to be aware of, is that “having that knowledge in one’s own head” might be unrewarding in time. Namely, once high-quality AI translation in near-real-time becomes the social norm, Chinese people (and most other people on earth) will become much less tolerant of a foreigner trying to speak their language imperfectly. You can already see something similar if you go to countries like Netherlands or Sweden and try to speak the local language at anything short of a high B2-level: people will complain you’re wasting their time compared to just using English.
Maybe because these is an obvious benefit of physical exercises and that there are no health benefits to learn a new topic after having already exhausted your brain sitting +7 hours at work?
> there are no health benefits to learn a new topic after having already exhausted your brain sitting +7 hours at work
There are. Unless your work requires almost constant creative reiteration over your mistakes on new topics, learning new things even when exhausted is still healthy for your brain.
I've been learning Spanish and I find a big positive motivator is finding and practicing slang words and vulgar phrases. They usually have cultural roots and are highly contextual, so it requires a deeper understanding than just translation. I only speak them sparingly with my few native speaker friends, who find it hilarious when I inevitably use them incorrectly - or rarely get it exactly right.
Learning a language is still absolutely worth it. The feeling you get when finally being able to communicate with native speakers is not something you could achieve with technology. Sure if all you care about is exchanging information then translation technology does the job, but if you want to actually connect with people I believe you have to do the talking
I've been learning Russian by myself since the pandemic and along the way I've picked up bits and bobs about food, history, geography, music, and even electronics. Learning a language is not just about the language per se.
I've even realised a few things about my own language.
It is helpful in such cases to look up, touch grass, and realise that "do it for you" is doing a lot of work there. The technology still can only emit a convolution of its training, and this is an ontological, conceptual limit on the technology, not something that the next model will just overcome. It's not "intelligence" -- you still have to know things.
It's easy to think, reading HN, that we're in some "post-knowledge" apocalypse, but that's just not the reality. It is, however, tragic that the irrationality of capitalism can be sustained so long, perhaps longer than some of us can stay solvent.
