Corvos
6d ago
As Sir Humphrey once said, "It is necessary to get completely behind a man if you wish to stab him in the back". The point is that people will resist the open implication, especially if it comes with apparent malice, but if you keep the relevant data points in front of their face long enough they will come to the conclusion you want themselves.
Corvos
6d ago
That’s good. Never pick a name you can’t scrub the floor in, and never pick a name that makes you look like a try-hard. If after a year you decide that the anglicisation isn’t enough, then change.
As a metaphor take cosmetic surgery - one should always get somebody else to decide the procedure or one just ends up with a face structured around your insecurities.
Corvos
6d ago
No, well-dressed men in their seventies who remember the Cold War and feel strongly about the necessity of Standing Up To Russia.
Corvos
6d ago
It’s a guilt thing. By not using the ‘correct’ non-Russian name he was letting down the Ukrainian war effort. Men of a certain age feel really strongly about this.
There's no getting around the fact that I use LLMs near-constantly, find them incredibly useful, and have considerable expectation for future use cases once the bottleneckers like Apple get off their arses. You don't.
I have old family friends whose ancestors felt that electricity was a fad and literally bankrupted themselves buying up candle factories 'cheap' in anticipation of renewed demand. My own grandmother felt that heated houses were fripperies and didn't get central heating until her 90s. I give these examples not to call you a luddite but just to say that literally neither of us knows yet and good, sensible people have been skeptical of in-hindsight very useful technologies (and vice-versa of course). My preference would be that we leave LLMs 20 years and see how it looks.
Corvos
6d ago
No, I get that, I don't like the marketing campaigns. But at some point you have to pay attention to what you're doing. If you're putting up flyers with a QR code, why haven't you scanned them at least once to know they work? And if you find they don't work, and I'm guessing 'work' here means 'link to your website' then why don't you ask the AI why they're not doing that? And if you've got no idea what you want the QR code to do, why are you asking for a QR code?
Managing AI is very much like managing humans. In both cases half the job is clarifying your thinking and soliciting follow-up questions and checking you're on the same page all the time, and "Well, I know what I want," doesn't work for AI or humans.
Corvos
6d ago
We could well be, I think some of the claims made are silly.
people aren't going to have to do anything and the market will eventually sort itself out. Supply will come to match demand
This is my hope! My concern is people turning it into a nuclear-power situation, where a perfectly good thing is loaded down with so much taboo and fear and regulation that it's no longer accessible even when everybody calms down.
Corvos
7d ago
The marketing for these things has been terrible from the start, I agree. Solving stupid 'problems' in stupid ways, stoking needless terror and antagonism...
And so unnecessary, too. These things percolated perfectly well with no marketing budget whatsoever.
Corvos
7d ago
In most cases one just gets increased convenience with evident downsides as well.
Just speaking from experience, the increased convenience is a really big thing. I'm having trouble with my employer in a way that's very time-sensitive, and being able to check their work while my lawyer was away on summer holiday was a life saver. I cannot stress enough how much it helped me.
I'm not opposed to this in theory but this feels like the wrong time: you can't do a cost-benefit analysis when neither the costs nor the benefits are readily apparent.
More generally... the conversation around AI feels like a Culture War conversation between lovers and haters squabbling over the pool of haven't-trieds. Both sides seem to have very strong confirmation biases; I'm a member of the lover faction myself and the 'I downloaded it, it told me a square has six sides, I deleted it' stories that haters tell me sound like they come from a different world which is a pretty good indicator that we're not operating on good epistemics here. If we're all operating on Conflict Theory then I don't think a big public discussion would produce very good results.
Corvos
7d ago
a significant fraction of the time it's literally on the level of "Please design me a rectangle." "Ok, here's that triangle you asked for."
I will say I used free GPT for a couple of years and never got this. But also:
the free version on mobile at least doesn't ask follow-up questions
How hard is it to put “please ask any relevant follow-up questions” in your prompt? This isn’t wizardry. We were doing this is 2023. Yes, you have to put in a modicum of effort.
And if you have an ounce of curiosity in you then you should really ask it ‘what is a QR code and how does it work’ when you say you want a QR code on your flier!
very little has actually come of this supposed improvement to coding. We haven't seen a flowering of new programs, and neither have existing programs gotten better.
Partly it's happening, just slower than you think and concentrated on business-to-business because these days building software for users has very high expectations for ease of use and frankly it's not worth it.
But the main issue is regulation and incumbents sitting on top of the bottlenecks. Want an app on your phone to point out you've missed an email that's urgent? Tough, only Apple gets to see your emails. Want it to be able to point out when you're doomscrolling? Tough, Apple doesn't let apps see what you're doing. Want your phone to take care of some trivial tasks without you? Tough, Apple definitely doesn't allow apps to use other apps. There's good security reasons for this but it means massive swathes of AI use cases can literally only be looked at by Apple and Google, who are slow-moving and mostly uninterested because they already own the market.
And this extends all over the place. Want to help people with something medical? I hope you've got a twenty person compliance team and you're friends with the regulators, and that's before you have a clue if your putative product will be popular or profitable. Similar for legal, similar for finance.
In short, the number of people who can do interesting things that you @OliveTapenade might notice in your personal life are limited to about 1-10 companies, who mostly have other priorities.
So far the primary use case for them is coding, right?
I also think this is too narrow. I've had a dispute with my employer and having something that can not only go through thirty pages of legal documents immediately and spot traps but also provide a legal fix so I don't have to rely purely on their drafting has been incredibly useful.
In 1980 you could say the same about the PC. When I was growing up, you could say the same about the internet. Lots of people's only interaction with the internet these days is Youtube and Facebook, because you need compelling content and incredible simplicity to interest the lower-than-average user, but that's enough for them to care about it now. Took ages.
Not that I ever saw. Reading about cool battles, yes. What are you thinking of?
Corvos
11d ago
Relevantly, literally no Argentinians have ever set foot on the Falklands. They were uninhabited before the British arrived.
Corvos
13d ago
Understand that there are broadly two types of programmer: the people who like making computer do things, and the people who like writing code because they enjoy logic puzzles and symbol manipulation. For the former (myself included), AI is a godsend. For the latter, it's an apocalypse.
Corvos
13d ago
I'm not sure I agree with the point it makes, but that's a really cool analogy.
When even the Chinese Communist Party says, “whoah, there, you’re creepin’ a bit”, I’m not sure that this is an unsparing look at how life really is as opposed to the fantasy of a rather nasty person who likes the aesthetic of power.
It's a possibility, though I hope not. One part of me thinks that it would be better if LLMs were never mentioned again outside technical documents - the public would be deprived of something that I think is an immensely important resource that could edify them, but on the other hand we would be safe from the kind of spasms of public morality that caused the censorship of social media.
The other part of me thinks that this is a temporary moment, where LLMs are used only by the most curious, and is looking forward to seeing what a generation of people now in their teens do with the tech, in the same way we saw in the flowering of internet culture 2000-2010.
Then we have very little quarrel. I broadly agree with what you've said, and I think the LLM marketers and grifters are doing more to damage the field than a thousand 'can't spell STRAWBERRY' bugs.
Really not my intention. This anonymous forum is where I can be totally open about my feelings because frankly none of this matters. I was trying to provide examples of cases where good, honest, clever people looked at something, concluded correctly that they had no personal interest in it, and concluded wrongly that it had no long-term value. I think it's an easy trap to fall into, and I feel the same in other areas.
In all sincerity, I've got my fingers crossed.
More options
Context Copy link