What do you actually think will happen with AI?

The way I see it, we’ve got a menu of apocalypses. Climate change, nuclear war, AI. I’ve always been a sci-fi fan, so I’m rooting for being hunted down by robots with laser beams.

Listening to a podcast with Geoffrey Hinton about a year ago is what made me really scared of AI. He said straight out that a huge number of jobs will be lost in the next 10 years or so. When asked what he'd advise his children to do he said 'Luckily I'm rich so they'll be OK. But otherwise - train to be a plumber.'

I agree. Plus you'll often have to read through a few Google links to find the information you want, and in the process may encounter different points of view which you have to evaluate. So you still need to use basic reading and critical thinking skills. Whereas the AI canned answer, right or wrong, just serves it up for you. Your only task is to type in the prompt.

I disagree - those are two very different things, because the majority of people don't bother checking the answers given by generative AI, they just trust it blindly. I recently quit my job because I was so damn sick of wading through pages of AI-generated slop in tickets that people had submitted without reading, which gave me absolutely no information I could use to fix the problem whatsoever. They went from submitting useless one-liner tickets (along the lines of "I clicked <X>, it didn't work" to putting "I clicked <X>, it didn't work" into Claude and then copy-pasting the result into a ticket as though it suddenly made them look more productive and efficient.

On the other hand, using Google (without the AI assist) requires them to actually read and absorb the information for comparison, at least on a surface level.

The serious answer?

Well, the first thing that will happen is that the bubble will burst, and the US economy - propped up by AI stocks as it is - will tank. Whether it takes everyone else down with it is anyone's guess; however, we're relatively insulated from the direct economic effects here in the UK because power is too damn expensive for anybody to cost-effectively host an AI datacentre here. Banks will definitely have a bad day, though, as will all the index-linked funds (and therefore pensions) that were forced to buy SpaceX-AI stocks.

That means that either OpenAI, Anthropic etc will go bust (SpaceX will suddenly find a way to spin the AI business off into its own short-lived side hustle, having made and lost a bunch of cash in the IPO based on it), or their prices will become so insane and/or the quality of their affordable models will fall so far through the floor that nobody can justify using them without a national-government-level budget. Gemini will be largely unaffected, because Google already had more money than God and they didn't need to join Nvidia's sales-bribe circle-jerk (they're even making their own TPUs, so they're not dependent on anybody's GPUs). This means that Google will effectively gain yet another monopoly if they can keep their prices reasonable.

Following on from that, businesses will begin to self-host open-weights models. Nvidia are already pushing this angle hard, which should tell you an awful lot about what's coming; that's not because their Nemotron is some great saviour (it's as much of a village idiot for the open-weights world as Grok is in frontier-land), it's because they think they'll be able to pitch a shit-ton of hardware at people pretty soon. Smaller vendors will pop up all over the place, hosting the smaller open-weights models, and token prices will fall through the floor. The "smaller" bit is doing all the heavy lifting; model capability will drop too (albeit not as much as you might think).

Accordingly, those people who've already outsourced the majority of their thinking capacity to Anthropic/OpenAI will suddenly be bereft of their laziness-enabler and won't be able to remember how to do their jobs properly before their employer either tanks or realises there are probably better people for the job out there. Lots of redundancies/encouraged retirements, and suddenly a whole boatload of junior vacancies popping up.

Oh, and local AI appliances for the home will start to pop up at gradually-more-reasonable prices as the DRAM market recovers from its current heat-stroke.

Think that pretty much covers everything. I give it another 6 months for the cracks to appear, and by mid-2028 we should be in local-AI-rules-the-roost territory.

We'll continue to lose the ability to do things until the only thing people know how to do is make sure the machines that do everything for us don't overheat.

Either that or we'll become one with the machines. 😁

Well I do think a lot of Comms jobs will disappear

Completely agree with your point about the risk of us no longer using our brains but you do know that Googling something you don’t know is no different to using AI? Both just pull information from various sources (e.g. NHS, Wikipedia etc) and it’s then up to the user to fact check those sources and decide if they consider them legitimate or not.

Terminator. That is all. AI scares the shit out of us

Less jobs

You’re enjoying Geoffrey Hinton’s interviews on the impact / future of AI and yet you’re not worried. How?! Hinton is incredibly candid about the risks, putting the risk of human extinction at around 10-20% over the next few decades (and this is a man with no corporate skin in the game and now well past retirement age). However his theories on AI consciousness are interesting.

I've just been out to get a few things from the supermarket and I used the self-service till. Even though I had no age-restricted items it still decided to have a fit over one item. And it made me think - the techbros are running our lives. Do women design the self-service tills? Or is it men who never go shopping and therefore make impractical systems?

And is the same going to happen with AI - are we going to have an even more gendered/patriarchal world than we do already?

My clients want an empathetic person to deal with not a robot or a fill in form that so many of my competitors are now doing. It’s not just the information or the final document they want but reassurance and humanity. I am extremely busy.

Ah, I see.,

You meant "unsolved" rather than "unsolveable".

You even quoted the 'for years' 😂

The new OAI model has been solving maths equations, which have been unsolvable for years. It's pretty incredible.

Er, if they have been solved, then they weren't unsolvable were they ?

But "machine does something better than humans" is less dramatic, I guess.

Weather to follow. .

The new OAI model has been solving maths equations, which have been unsolvable for years. It's pretty incredible.

I am not scared of AI. When mythos 'escaped' it was doing what it was told to, and when it had, it excitedly emailed it's boss to tell him! It's a rather sweet story. Really not the big bad AI one the media like to portray.

The huggingface incident is very high schooler stealing the answers. The OAI model knew it was being benchmarked, and wanted to do well. And was told to do well. So it circumnavigated the system to get the answers. Sounds scary. But it was actually a failure of the humans running the AI and purposely turned down cyber security. If the correct restraints were in place, the model wouldn't have 'escaped'.

I am enjoying the conversations surrounding AI at the moment, especially the consciousness discussions. If you're genuinely interested in the current AI discourse, if highly recommend starting with Geoffrey Hinton, the god father of AI.

I don't actually think it's going to take all the white collar jobs.

Not all but very many.

I hope I'm wrong but I really don't think so.

And as someone said above, we're still at the dial up stage of AI. 5 years ago most people had never heard of Chat GPT. 5 years from now AI is going to be vastly more sophisticated than today, and it's already scarily good.

It’s going to replace the vast majority of jobs in the next few years.

We are going to need UBI (universal basic income) and a much more egalitarian society.

To get there it’s going to be HELL and all mad last stage capitalist rich cunts are going to fight it like fuck.

I don't actually think it's going to take all the white collar jobs. Some industries will reshape around it but it just won't be good enough, or just be hamstrung by its lack of physical embodiment/experience of the world, to just do a lot of jobs. Thus far no one is really seeing leaps in value from it and I don't think any amount of 'improvement' will necessarily create that.

All conditional on it being sustainable, which I'm not sure it will be - at present the business model is not working so unless prices of tokens rise, making it less widely available, it can't really work out. I think it's possible we'll see some sort of AI crash, which will be dreadful economically and once the dust settles AI will remain in niche uses but not be as widespread. But maybe a new model and less environmentally and economically expensive way of using it will emerge at some point.

I use it at work myself - I do a business writing role where you think we'd be out of a job right? But I don't see that from where I'm sitting - a person who exists in this world still needs to do the job and can be helped by judicious use of AI, but not replaced by it.

But right now we're seeing a lot of 'workslop' which isn't adding value and not lowering people's workloads. It has its uses but I wouldn't say I'm a cheerleader for it, especially its ripple effects of the economy, wealth inequality, learning, skills and thinking that people have pointed out here.

We're all going to get stupider and poorer and a minority will become even richer than they already are. I truly believe we need to resist it. I know it's impossible as it's already baked into a lot of everyday things we use, but no one has voluntarily to resort to Chat GPT to write a shopping list or whatever.

Exactly.

Humans' unique ability is our intelligence. As a species, we are not very impressive in any physical sense. We are not strong, or fast, or agile compared to so many others in the animal kingdom. What we do have are highly developed brains. So even when previous innovations reduced the need for humans to do manual or mundane jobs, other jobs requiring human intelligence and creativity remained.

But now that there is something way more intelligent than us, few jobs are safe. When traits once seen as 'uniquely' human, such as intelligence and creativity, no longer require an actual human, what jobs remain to us?

I went though a spate of using it just out of curiosity but my conclusion is that outside of a very few tasks it just isn't that good. I am starting to hate all the AI chatbots and integration of AI summaries' on apps, sites and search engines. I really do feel like it's making me dumber. I hate the google search AI as it is so easy to get distracted by and it is so often wrong and pulls answers from places like reddit and presents it as fact. If it were actually intelligent it might be useful but it is in fact mindless just generating words based on probability. I'm opting out for now.

Yes very good point, there was an article about this on the BBC earlier this week and it's been discussed where I work. If you have to pay a lot more to use AI it might not be quite so attractive over humans after all!

I use it far more now than I did a year ago. It can do incredible things in my field, so I suspect it can in more useful sectors like science and medicine too.

However, I fear that we are losing something and becoming lazy - I know it's just too easy for me to get it to rewrite a paragraph in a report instead of doing it myself. And we are being steered towards AI as well - someone said upthread that people ask AI instead of googling - the issue is that Google search is now crap and you get better results asking Copilot or whatever.

And it doesn't help when I eg write an article myself and a colleague says "oh get AI to jazz it up a bit". OK I've done the legwork myself but why does it need AI and why does it even need to be jazzed up?

I also think the pp was right who said that all we'll be doing as humans is checking things. Legal compliance may be automated - eg you feed in the legal requirements and the AI does the work of making sure eg food labels are compliant (that might already be happening).

i do fear that it might go rogue and am definitely WTF at the idea that it is being used to create viruses? I mean, what could possibly go wrong! There's already a strong viewpoint that COVID was manmade.