No, I absolutely realise that: but it's irrelevant. It doesn't make them safe.
And after reading the linked BBC article, along with the article linked from the discussion at https://news.ycombinator.com/item?id=48997548, I struggle to grasp the failure of imagination that doesn't see or accept that going further and allowing an LLM to control something that acts physically in the real world could be incredibly dangerous.
It really doesn't matter whether that behaviour results from "intelligence" or "awareness" or merely a stochastic quirk of a particular run of the model(s) and their orchestration: a dangerous machine is still dangerous.
It's odd that you would try to dismiss that concern with a non-argument.
> was being tested in a controlled environment, but found vulnerabilities and managed to escape.
Uh-huh.
It's worth keeping in mind that we are here talking about people who simply will not be satisfied until they have built Skynet[0]. They will then no doubt experience some very brief satisfaction before we are all annihilated.
I can't help thinking the juice isn't worth the squeeze.
I'm not sure what the alternative is or how to change course unless and until it becomes unambiguously apparent that such a thing is not possible. Currently it seems like we still think it might be possible, so that's where we're heading because if "we" don't do it, somebody else will.
[0] How feasible this really is, or on what timescale it might be possible, I don't think anyone can really say. But, at least for now, this is clearly the aim and trajectory we are on.
As someone who watches a reasonable amount of PoV outdoor activity footage shot on helmet cams and the like (base jumping, mountain biking, skiing, snowboarding, etc)… I don’t love watching 360 videos uploaded in the raw because of the perspective distortion.
I’m assuming it must be possible, if the resolution is good enough, to post process a portion of each overall frame into an undistorted 1080p (or better) view of the key view of the action, but a lot of people don’t do this (perhaps it’s much more difficult or time-consuming that I’m imagining, or perhaps many viewers enjoy the distorted 360 view more than I do).
Yeah, we ordered a double width pax with sliding doors, drawers, shelves and lighting, plus another half wardrobe with a mirror door, shelves, and lighting.
I’m fairly handy and quite enjoy putting things together so it never crossed my mind to use IKEA’s assembly service, but I did notice they were offering it for £300, which I took as a signal that it would take a while.
In the end it took 12+ hours over the course of a weekend. I reckon I’d be faster if I did it again, and maybe if I did it enough times I’d be able to half that, but the problem is, how many fully spec Ed pax wardrobes + ancillaries are you really going to assemble in your life?
With bigger orders, depending on how you value your time and want to spend it, maybe paying for assembly makes sense. I reckon 2 people would have had it done in a morning if I’d paid.
I just noticed that by around 2015 it had got super toxic and snippy.
Usually I’d find answers on SO. Relatively rarely I’d ask questions but, when I did, I’d always try and follow the netiquette rules of yore, and think in terms of, if I was a support engineer trying to help with this, what would I need to know?
Because I have supported products, and we’ve all seen enough bug reports and questions come in that we can tell when someone is going to be easy to help - even if they have a particularly tricky problem - versus someone who’s going to prove more challenging.
So I had this question about Elasticsearch, and it was at a time when the documentation wasn’t great, and you were actively encouraged to go on SO and tag your question to get help.
I wrote out in detail what I’d done, where I’d got stuck, what I’d read and tried to get unstuck, etc. It probably took me 30 minutes or more to pull everything together into a coherent post.
The very first comment was from some insufferable bellend saying, “Oh, so you want us to do your work for you, are you going to pay us too?” or words very much to that effect.
Literally, WTF? Why even post that? If you don’t want to help the option to simply go away without getting involved is always available.
IIRC I didn’t actually end up finding a solution via SO and instead layered some godawful hack on top of Elasticsearch to get what we needed - because I simply had other work to move on to and I’d already spent a lot of time on the problem.
But I think that was the last question I posted on SO, and maybe the last time I posted anything on the site.
As the years wore on I simply started finding it less and less useful, with often incorrect answers marked as accepted and - if you were lucky - the correct answers marked might be buried further down.
And then there’s what they wanted to charge for job ads versus how effective those ads actually were - again, this was better in their earlier years.
SO started out well - genuinely a breath of fresh air - but as time went on it felt like they thought their model was the last word in online help forums and they didn’t want to evolve to address its flaws, even if that had just been dealing with the toxicity, and the karma farming.
And so this is the result - a site that, like the dinosaur in A Sound of Thunder, is dead but perhaps hasn’t realised it yet - and, at this point, the way I feel is simply good riddance. It’s a shame, but - as you said - they did it to themselves.
It’s also practicality and ergonomics: old furniture is nice to look at but it’s often awkward to use (e.g., squeaky wooden drawer runners that require periodic waxing to keep them running smoothly and quietly, lack of sliding doors, dark corners at the back of wardrobes, etc.), and wastes lots of internal space.
Drawers with wheels and bearings and soft closers on their runners are simply a lot nicer to use on a day to day basis, particularly when they’re used frequently.
I do have quite a strong preference for older furniture from an aesthetic perspective, and we certainly have some older pieces as a result. But it’s often functionally deficient, and so we don’t tend to use older furniture more complex than a table where the functionality is frequently used. In particular, any kind of heavily accessed storage means ergonomics and usability win every time, tending to steer us toward modern options.
Sure, it’s not hard, but if there are a lot of steps it’s still going to take longer.
A fully specced up pax double wardrobe with drawers, sliding doors, and lighting is made up of hundreds of parts, so does take a good amount of time to put together even though it’s not difficult per se.
It’s also made up of many different SKUs so you get dozens of boxes turn up and it takes a little while to figure out what everything is and get organised. Again, it’s not difficult, but it takes a little time.
Weekend HN is definitely where the more interesting and offbeat content lives, and I often find myself enjoying it significantly more as a result.
During the week, well, it would be unfair to call it LinkedInified, but it can often feel like a somewhat higher tier of that sort of strata. Plenty of good stuff in there still, but much more “serious business”.
One does find oneself slightly askance at one's own thinking sometimes, that's for sure.
But I suppose, is it really so different? I mean, back in the day moreso than now, a lot of the valuable IP in any system was in the design and specification of that system - the problems usually solved within the design and specificaion (use X algorithm, etc.) - and the code was "just" the implementation of those solutions.
So perhaps it's more of a regression in some ways: the value is in the specification (the prompt) once again.
Your point about stochastic behaviour is well made though, and there is no way to 100% guarantee or formally verify the behaviour of a system that relies on an underlying technology whose behaviour is fundamentally stochastic.
I don't know how you'd enforce that unless it was something you could mandate at the level of the API call, and then the API call is rejected if the customer hasn't configured it for "no transcript".
It sort of feels like an area of friction even still.
I imagine this will be because a decent chunk of the IP in Codex is probably within its prompts, how they're built, and how they're sequenced and orchestrated, rather than in the codebase per se.
We had this discussion a few months ago where we talked about allowing people to choose an AI provider and provide their API key, thinking about enterprises with "preferred" (read: mandated) AI suppliers. We also wanted to offer the kind of very simple pricing that this is one way of enabling. But we realised pretty quickly that this would/could lead to leaking our back end prompts to customers and, although those prompts are only a part of the value add, if you could build a detailed trace of them then you'd be able to relatively easily reverse engineer a lot of what we're doing.
Or people screw up the placeholder content and call you by the wrong name, wrong job title, wrong company, whatever (off by 1 errors in some columns in their automation sheet?).
It's already happening with outreach messages on LinkedIn. Gets an instant block.
Yeah, I think this is it. The humans were the point, not the minutiae of the tech.
(Btw, I hadn't noticed you'd responded whilst I was editing my comment to express myself a bit more clearly - I hope anyway - so the quotes don't quite match but I don't think it matters, because the sentiment is hopefully clear enough both ways!)
It's funny... I enjoyed Neuromancer, although I didn't read it until about 15 years ago.
And, yeah, the memory thing hasn't aged well. Thing is, 1984 was a funny time in computing, particularly when you consider the kind of computers normal people had access to.
At that point even things like PCs and the new Mac had 128 or 256K of RAM[0], so I get that 3MB must have seemed like an ocean of memory at the time. And, realistically, more than 1MB of RAM in machines you'd typically see sat at home or on a desktop was uncommon until the beginning of the 1990s.
And, although Moore's law had been around since 1965 it's hard to know how aware people outside of specialist circles would have been of it in 1984.
I suppose Gibson must have done some pretty in depth research for Neuromancer, right? But the memory thing is sort of ancillary to the story, so how much would he really have focussed on that? Probably not much.
And then do you really want to harshly judge the book on that one slightly laughable thing, in other ways, it was incredibly forward looking and almost prophetic? Doesn't seem right.
I think the sensible position is you have to let it slide and see it as a possible alternative future that never quite came to pass in that way but that which we can see strong echoes and foreshadowings of even still.
[0] In 1984 microcomputers, as opposed to, cough, "serious" computers like the PC and Mac, with 128K of RAM were still very new, with 32 - 64K being the entry level, and if you had one with 128K you were king of the hill. 128K in 1984 seemed like a ton of memory to most of us, but it's worth bearing in mind that only a handful of years before computers like the ZX81, which had only 1K of RAM, were the common entry level, so the progression was already clear if you looked at the situation in the right way, but you had to have been paying attention for a while to have noticed. I remember the first time I used a machine with 4MB of RAM in, maybe, 1990 - an Archimedes at school - and feeling like it was just this absolutely inexhaustible ocean of memory. In 1984 3MB would have felt almost inconceivably huge unless you were in the high performance computing, or maybe the mainframe, worlds.
Whoa, wait, what? It can? I did not know that. Back in the day I had an A500, and only bought the 1200 decades later. Had no idea it would accept HD floppies. TIL. Thank you!
Yeah, but eventually the downvoting/community moderation will do its work: it just takes a few hours sometimes. But the original unkind remark has gone from top level comment to nowhere near the top of the discussion (at least I haven't managed to find it other than via my own profile because of all the newer and more upvoted discussion that's happened) fairly quickly.
I'm not talking about you though, am I? I'm saying that there are people working in this field who will do exactly that.
We are now going in circles.