Is it worth trying to resist AI?

[RichardT]

We all know AI is coming, and it will be very capable.

Hollywood actors and writers are trying to resist it by striking - they want it to be regulated or blocked to protect human jobs.

I've tried looking on the web to understand whether there have been any times when resisting a technology has worked in the past - either to stop it altogether, or control it.

It seems there are lots of cases where regulation has been put in place for new technologies, such as cars and television, aiming to limit their success, but it seems in these cases that eventually the technology became ubiquitous anyway.

The same seems to be true for older technologies such as power looms, steam trains and the telephone.

There was even sabotage of computers in the 1970s in the US by the Computer Liberation Front.

These attempts all seem a bit quixotic in hindsight.

Have there been any successful efforts to do this where the technology has stayed blocked or controlled? I haven't found any so far.

The closest I've got is nuclear technology, because it's potentially very dangerous.

[Albatross]

I think big oil and the motor manufacturers have likely buried technologies that threaten their grip... electric cars, hydrogen... all been around for years but never given enough oxygen to get going. Tesla (the man) with his free energy distribution systems. Loads of stuff.

You just need something that's worth more and has more vested interests than the new competitor; which will then be bought out, discredited and buried.

But with AI I reckon its out of the box and everywhere already... not a chance of stopping it.

[James Perrett]

The AI engines are freely available and the Internet is a great source of training material for many (but not all) purposes. There's nothing to stop a kid in their bedroom from creating the next must have AI tool.

[Drew Stephenson]

I'm minded of the quote I came across a few weeks ago, "AI won't take your job; but the person who knows how to use it will."
It's like most tools, some use will be good, some will be bad, some will doubtless be horrific, and a few things will turn out beautifully. But all of that will be because of the choices made by humans.

Is it worth resisting? I doubt it.
Is it worth having serious discussions about the ethics, impact and potential regulation? Absolutely, and the sooner the better.
But that requires honesty, informed opinion and a genuine set of motivations. We're a way away from that at the moment sadly and most of what we see is either grandstanding, political positioning, or doom-mongering.

But this is also a repeatable and predictable cycle.

[RichardT]

Yes, we're in the horror and ignorance phase now, as per the pandemic.

I think regulation is desperately needed to stop AIs being used for nefarious purposes, but I'm afraid it would be relatively easy (though expensive) for criminal networks to build their own. Casual nefarious use of the major AIs could be blocked by the software itself, though.

Based on past examples, it seems that regulation of use, if it were possible, could reduce the speed AI takes off, but only for a while. It might help ease the transition. But in the current globalised world, I can't see it happening.

Ban AIs in the US from writing scripts, and someone in China or India will do it anyway

[muzines]

See also: Most disruptive technologies - steam power, the loom, the printing press, radio etc...

[Drew Stephenson]

Yep, I don't think banning is the right answer (but then I rarely do so there's that old confirmation bias). I think it's about focusing on understanding, communicating the benefits and risks, and then regulating for extreme uses.

I find it stupidly frustrating that of all the things we could be trying to get AI to do we seem to have started with the things that humans most enjoy doing. Why not start with tax-returns and digital assistants for labyrinthine government form-filling?

But like most of these things, we'll figure it out. Ultimately we in the west generally work fewer hours at less arduous and less dangerous jobs than previous generations did (occasional blips aside).
Yes it's enraging that a tiny percentage of people have disproportionally benefitted from most of the wealth generation of the last forty years (really that's on us for voting for politicians who've enabled it) but in time this might be the next thing to bring us closer to the four-day (or less) working week.

But there will definitely be times along the way that will be very ugly.

[amanise]

One thing we can be sure of is that AI interests will be directly aligned with monetary interests. I'm sure they already are. Those with the most money - will be able to commission and 'control' the most AI. All you need to do is look around the world at who is already super rich to see whose interests are going to be first and most served by the first 100 years of AI.

I sometimes wonder if AIs will then start to derive their own systems of ethics and regulate against us - but then there have been many musings around that already.

[OneWorld]

I find it stupidly frustrating that of all the things we could be trying to get AI to do we seem to have started with the things that humans most enjoy doing. Why not start with tax-returns and digital assistants for labyrinthine government form-filling?

Spot on

[OneWorld]

I find it stupidly frustrating that of all the things we could be trying to get AI to do we seem to have started with the things that humans most enjoy doing. Why not start with tax-returns and digital assistants for labyrinthine government form-filling?

Spot on

[OneWorld]

"Is it worth trying to resist AI?"

I suppose the same sort of question was asked when man/woman/whatever discovered fire or when the wheel was invented - we sort of got used to it.

But when you atomise an issue, boil it down to the basic elements, nothing much will change, we'll still go weak at the knees when the gas/electric/broadband/water/insurance/mortgage etc bills come each month, except AI will be employed to find more effective ways of emptying our pockets.

And almost any technology does nothing except change parameters, whether it be a horse and cart, a steam engine, an internal combustion engine or an electric motor, essentially the endeavour remains the same - to get from A to B and the rich will get ever richer by convincing us we have to buy into the latest fad. Yes the convenience issue has to be factored in, but come what may, the sun will still rise in the East tomorrow and if we tighten our belts, governments will tell us sunny uplands and wealth beyuond compare are just around the corner.

Who is resisting AI anyway?

[OneWorld]

We all know AI is coming, and it will be very capable.

Hollywood actors and writers are trying to resist it by striking - they want it to be regulated or blocked to protect human jobs.

There was even sabotage of computers in the 1970s in the US by the Computer Liberation Front.

"Hollywood actors and writers are trying to resist it by striking - they want it to be regulated or blocked to protect human jobs"

Thing is, it was seen as bringing big changes when the TV broadcast license restrictions were freed up and instead of just 5 channels we got 55 (or whatever) but 5 channels or 55, 99% of the content is still bobbins. There is no way Hollywood can knock out great content 24 hours a day, 7 days a week, 12 months a year. Same with AI, yes things like chatGPT can scrape data mines, but 99% of that data is hyped up bobbins anyway, unless you have some specific interest in The Sex Life of a Congolese Dung Beetle

It is understandable thespians and writers want to protect there jobs, but jobs will go, just as blacksmiths' jobs went when the motorised perambulator was invented. What makes the human race the most successful species on earth is our adaptability, thing is we are creatures of habit and don't like chenge so change is often forced upon us, and then when push comes to shove, once we have changed, we like the change, despite our complaining. It's like the lions and wilderbeest on the plains of the Serengeti - some see challenges and some see opportunities

"There was even sabotage of computers in the 1970s in the US by the Computer Liberation Front."

Yep, and there are still some people that say the earth is flat - well it is to a flatfish I guess, but not to a spaceman

[OneWorld]

Ban AIs in the US from writing scripts, and someone in China or India will do it anyway

Well why wouldn't they, write AI scripts in China or India, they already have, or Nicaragua or Nauru, or the Vatican City etc where do we get this notion that the USA holds all the aces and has domain over everything everywhere?

[Drew Stephenson]

Interesting piece on Techdirt yesterday, less about the 'will it / won't it' of AI but why people, particularly in the creative industry, keep turning to copyright as a tool to try and protect themselves. https://www.techdirt.com/2023/07/19/sto ... ems-of-ai/
Copyright and patents are literally there to promote progress in 'science and the useful arts'.
There will be regulatory tools developed that help us to use it, in the same way that there was for the motor car or the aeroplane or the internet.
But they won't be driven by copyright.

[merlyn]

The flaw in that is treating an AI like a person. Yes, people read books then write one, and that's generally agreed not to violate copyright. Photocopying a book does violate copyright. Photocopying uses a machine that can't read to reproduce a book. An AI is a machine, not a person.

[Drew Stephenson]

Indeed, in some ways it's a weak analogy, but my point was that using a law that was written to promote progress to try to hinder progress isn't likely to end well. For anyone other than copyright lawyers.

[Folderol]

This might be of interest:
https://www.lightbluetouchpaper.org/202 ... n-exhaust/

[Drew Stephenson]

This might be of interest:
https://www.lightbluetouchpaper.org/202 ... n-exhaust/

Aren't most of them set to only read content that's pre-2021 for exactly this reason?

[Folderol]

This might be of interest:
https://www.lightbluetouchpaper.org/202 ... n-exhaust/

Aren't most of them set to only read content that's pre-2021 for exactly this reason?

Possibly not. I've read in a number of places that the 'quality' is degrading.

[RichardT]

This might be of interest:
https://www.lightbluetouchpaper.org/202 ... n-exhaust/

Aren't most of them set to only read content that's pre-2021 for exactly this reason?

Possibly not. I've read in a number of places that the 'quality' is degrading.

AI generated text has a very distinctive style - news sites are already using it to generate their copy, such as this one

https://ts2.space/en/toyotas-revolution ... ile-range/

This makes me want to resist! At the very least the AI should be credited, so that there is some kind of openness about it.

If this sort of stuff gets fed back into the inputs then we are going to live in an epically bland world.

[Drew Stephenson]

Possibly not. I've read in a number of places that the 'quality' is degrading.

Interesting, and rather dumb.

This makes me want to resist! At the very least the AI should be credited, so that there is some kind of openness about it.

Strong agree.

[ajay_m]

Ok, imagine this. You are wearing a headset and working with your daw. You say "insert track. Insert plugin, kontakt. Load spitfire brass, trumpet ensemble. Insert FX, plate reverb, 1.4 sec.
Insert track, drums, bossanova beat, jazz kit
Or
Insert track, solina, reference Pink Floyd animals.
Or
Rewind. Solo track 1,2,3. Play. Volume up 3db.
Or
Insert channel strip. Reference Bob Clearmountain.

And so forth. AI could be an enormous time saver not a threat. In fact I'm sure someone's working on a voice activated control surface and this will be just one of many ai productivity aids. Removing reverb from a recording without artifacts. Properly dealing with fret noise or removing plosives or sibilants from vocal tracks automatically. And so on.

[RichardT]

Ok, imagine this. You are wearing a headset and working with your daw. You say "insert track. Insert plugin, kontakt. Load spitfire brass, trumpet ensemble. Insert FX, plate reverb, 1.4 sec.
Insert track, drums, bossanova beat, jazz kit
Or
Insert track, solina, reference Pink Floyd animals.
Or
Rewind. Solo track 1,2,3. Play. Volume up 3db.
Or
Insert channel strip. Reference Bob Clearmountain.

And so forth. AI could be an enormous time saver not a threat. In fact I'm sure someone's working on a voice activated control surface and this will be just one of many ai productivity aids. Removing reverb from a recording without artifacts. Properly dealing with fret noise or removing plosives or sibilants from vocal tracks automatically. And so on.

Absolutely - eventually I would like to be able to say ‘sort out the dynamics on the piano part’ !

[Drew Stephenson]

Indeed, I'd settle in the meantime just for some voice activated stuff so I could be calling out what how I want the software set up whilst I rig the hardware.
Let's get the AI doing the drudge work and leave the fun stuff to people.

[Folderol]

If you no longer had to put any effort into it I think you'd fairly soon find it boring.