Forum menu
1-10 - with 1 being everything carries on largely the same with the aid of slightly useful assistance and 10 being the Terminator conclusion, how scared are you of AI right now?
I've started to pay more attention over the past month due to thinking (naively) about how I can be a bit more useful at work but flipping 'eck, after watching a few interviews with Altman, Schmidt, the AI2027 book its all getting a lot more concerning than being able to mess about with it on Excel! - combined with the fact that as humans we're not particularly good at handling issues of this magnitude in a good way.
Its the rapidness of the advance I think is the most of my issue, there seems to be a lot of high level people asking for legislation, laws, controls being put in place but its coming at such a slow pace compared to the advances of the technology, I guess one small hope is the energy needed to maintain development lags behind as we just don't have the capacity to service the development short term.
5-10 years I think the world will go through social-economic change hugely greater than the industrial revolution, it's alright saying I'll retrain as a plumber but when the countries awash with plumbers, sparkies, tilers etc the competition will drive prices to the bone and who'll actually have cash to spend bar a fortunate few?
be interested to get the take of others on this...
Its the rapidness of the advance
I think rather than the technological shift it I actually think the pandemic has been the accerating factor - it totally upended our view of work and worth. Turned out the lowest paid manual worker's work was 'essential' and the rest of us could just go home. I think it was the boss of John Lewis who said we saw 10 years worth if shift in retail habit during 2020 - in terms of how we think about cash, online/offline shopping and so on. I think we didn't really 'see' the shift in attitudes to work more generally and as a result AI technogolies have been able to slip into an already changed mindset. Its not a case of legislation etc being too slow to change, our habits and practices changed ahead of the technology and the technology is filling a void we've created.
If it turned out you could work from home then I'm afraid you might not have come out of that year of change as a winner, even if you might have felt like one at the time - if the work you do can be done without your physical presence - I think its only going to be out of nothing other than your employer's habit that'll keep you in that job and the doors are going to be closing for other jobs like it. There are already quite sizeable companies that require recruiters to prove a job couldn't be done by AI before they can even advertise it. If you can use AI to help you do your job then AI can do your job and it won't take long for your employer to realise they can just AI to do your job instead
It truly has the potential to cause serious issues but unfortunately not the fun terminator apocalypse more a David Brent office sink into gobbledygook.
In a drive to save money ai will be encouraged by managers who don't understand it's flaws. In it's work due to AI bias and hallucinations it's output will become more and more flawed until it sounds fine but is just nonsense. Think uncontrollable management speak but with less purpose. By this time it is "managers" using AI to generate output that is read by another AI but because it's easy and cheap no one will check the output too carefully.
It truly has the potential to cause serious issues but unfortunately not the fun terminator apocalypse more a David Brent office sink into gobbledygook.
In a drive to save money ai will be encouraged by managers who don't understand it's flaws. In it's work due to AI bias and hallucinations it's output will become more and more flawed until it sounds fine but is just nonsense. Think uncontrollable management speak but with less purpose. By this time it is "managers" using AI to generate output that is read by another AI but because it's easy and cheap no one will check the output too carefully.
TLDR - 3
Robots and computers have been about to take over all jobs since I left school in the early 1980's. The shift has been happening for a long time and has created many jobs as well.
I note everyone is jumping on the AI bandwagon as they often do with the latest 'buzz word'; yesterday I was in a mobile phone repair shop and they were selling "Designed for AI" Screen Protectors 🤨 .
I tried to use AI to tweak a Settlement Agreement recently and it produced utter garbage. I tried rewriting a short email in a different style, that was a bit better, but still contained fundamental flaws.
I think it is/will be good, but it's a long way from what people fear at the moment. The rate of progress is difficult to predict, but people like my son should potentially embrace it as they still have 40+ years of employment remaining.
yeah, so many questions, if you have the capacity to use ChatGTP in your role at work why keep the middleman...I'm sure we're all well prepared for some kind of universal income type society based on I've no idea what.
Having just watched a few videos showing Geoffrey Hintons thoughts I'm off outside and away from the screen, his take on 'how many instances does a less intelligent thing control a more intelligent thing' is not helping - combined with the tech being in the hands of a very small number of companies who may/may not be trying to win a race at all costs. May have to limit my screen time on this subject.
if the work you do can be done without your physical presence - I think its only going to be out of nothing other than your employer's habit that'll keep you in that job
People working from home are physically present - at their computer. Where that computer is located doesn't really matter in lots of cases.
Some jobs will go for sure but I don't really see all jobs that are knowledge based disappearing.
Large companies are already awash with bullshit jobs where it is hard to understand what the actual value of the work is. I am definitely seeing a shift in the type of bullshit being spouted, but it is still bullshit.
Agreed about the Pandemic. I've always done manual work , the last job before retirement was shelf stacking. It was funny the difference in how we were treated during COVID, suddenly shoppers started to notice us instead of us being invisible, shame it was only for a very brief moment 😔
I see the knowledge/technical roles going in some kind of second phase once theybe got most of the customer facing roles sorted via chat bots etc, I came in from a works angle and I have 5yr old twins that'll be going into a much different jobs market than todays, but the work was just the 'in'
I'm now much more worried about the existential threats, lack of controls, short term governance etc as in - where we're going we wont need jobs, or people.
Having just watched a few videos showing Geoffrey Hintons thoughts I'm off outside and away from the screen, his take on 'how many instances does a less intelligent thing control a more intelligent thing' is not helping - combined with the tech being in the hands of a very small number of companies who may/may not be trying to win a race at all costs. May have to limit my screen time on this subject.
I watched a good video from Hinton few days ago, gulp 🙁
Meanwhile on another thread
Not interested in anti Elon rants.
so when we end up as slaves to the tech oligarchs, don’t forget it started when we got dazzled by shiny toys.
Robots and computers have been about to take over all jobs since I left school in the early 1980's.
There's a difference with AI though, it's fundamentally non linear non sequential non algorithmic in how it operates. Thats the potential threat, that it can think better than us... it can create faster then us.
However it fundamentally runs on systems that are exactly all those things - linear sequential algorithmic. The layers of code complexity built up on the latter to create the former, alongside the enormity of the data sets it most process to be of any use at all are enough to require banks upon banks of the most powerful cutting edge technology arranged in data centres the size of football pitches.
It's very expensive to set up and also very resource intensive. Sorry have lost my train of thought... in danger if waffling!
I'm sure we're all well prepared for some kind of universal income type society based on I've no idea what.
Given what's going on in Gaza and how much resistance by our leaders to do anything about it, a universal income based society seems unlikely.
That's my main hope in all this, the US seem to be restricting chip access to China but they've shown very good ways of getting around that, it's then a power bottleneck which China is well in advance of the US on, can very much see a Nuclear style AI mutual assurance treaty between US and China in years to come.
flipping 'eck, after watching a few interviews with Altman, Schmidt, the AI2027 book its all getting a lot more concerning
I would caution you to take Sam Altman with a massive pinch of salt though.
OpenAI is the biggest corporate money pit that has ever existed, burning through billions every month with no realistic path to profitability any time soon.
The only way it can continue to exist beyond about the end of this year is if Altman can convince potential funders that it's going to change the world and they'll be left out if they don't sling him a few billion to get on the AI train.
So yeah, he is close to the action, but his interests are very much in hyping it up, rather than clear-headed analysis
Like a load of other people I use out to get my work done quicker/better. Board reports produced x10 faster, summary papers, project plans, job descriptions so done by Ai with me refining them, so using only 10% of previous time.
I’m currently supporting my wife with a business purchase and we are paying for traditional legal and accountancy due diligence. But I’m doing everything in parallel through AI and the outputs are just as good as what we are getting through paid services. If we were doing this again, and now with more knowledge of process, I reckon we could save £15-20k in fees by doing a lot of the early work ourselves through AI, if we had the confidence to do so.
Indeed, just a few examples from the last 10 years of hype from cash rich tech companies desperately seeking the next big thing:
Wearable devices
Autonomous vehicles
Blockchain
Cloud computing
Metaverse/VR
I still like people. I suppose not paying people appeals to lots of other people. But what’s the end result of that?
Regarding the infrastructure requirements of AI, I was thinking the other day just about YouTube.
Literally billions of users can access any point in any video within seconds.That is on top of the fact that millions of users are also uploading videos constantly.
It is quite impressive really, which lead me to realise how much that is taken for granted. It's not that long ago it would have been unthinkable on the this scale.
...it's fundamentally non linear non sequential non algorithmic in how it operates.
However it fundamentally runs on systems that are exactly all those things - linear sequential algorithmic.
This is an interesting point. Will the next generation of AI that runs on quantum computers be another level of sophistication/danger (delete applicable) than that we see today?
This is an interesting point. Will the next generation of AI that runs on quantum computers be another level of sophistication/danger (delete applicable) than that we see today?
Yes. Potentially increasing it's knowledge far beyond what we're capable of keeping up with, once the most knowledgeable brains in the world are asking chatGTP what's what it's a brave new world, although that does go on now with recent medical advances.
AI moving away from say English as a base and starting to use its own more efficient language that allows it to progress at a faster rate is just one of the guardrails that needs sorting. There's a lot of dooms day scenarios to sort
'They spent so much time thinking about if they could and not thinking if they should'
mm... scared as in scared of what it's going to lead to: 6-7. Microsoft is literally recommissioning Three Mile Island to power its servers for 'AI'; greenhouse gas emissions are spiking as a result of the power requirements. And, fundamentally, AI as it stands is dumb - it reorders words and information that has been fed into it, into forms that look like new writing and information, with no checks for veracity, rightness or appropriateness - but people think it's fantastically clever. It's going to lead to major problems/ military incidents/ economic meltdowns.
Scared as in it'll do me out of a job: 2. It ain't all that; by the time I come to retire it may be getting to a point that it can do parts of my job better than me.
by the time I come to retire it may be getting to a point that it can do parts of my job better than me.
There's an argument that we're at Peak AI in terms of current models - they've been trained on "the Internet before AI" and any generic AI that follows will be trained on "the Internet including a load of questionable content produced by dodgy AI". AI is proving to be particularly useful when it's trained on specific data sets with defined outcomes (eg finding cancers) but we've all seen how poor it can be when trying to cover a much broader knowledge base.
As for jobs - there have been many historical changes that have automated work that people used to do, but people adapt to the jobs that haven't yet been automated. I don't see a future where the machines do everything while we have 24/7 leisure time.
One thing that's worrying that I've noticed a few times now, watched a 'documantary' on youtube the other day, narrated by a male voice with the kind of 'middle american' accent you might expect for such a thing.
The then the narration said something along the lines of 'these birds live on the steep cliffs of the island'
But it pronounced 'live' as in 'live on radio 1', rather than 'live' as in lives in a house.
Makes you wonder how much else slips by that doesn't contain such obvious red flags, or factually wrong information, and people would just assume it's correct, as it genuinley sounded like a person doing the narration.
This group are terrified of AI, enough that they shot some border guards a while back, It's a pretty wild story. What most of us interact with online is mostly LLM rather than true GAI, but I think there should be internationally agreed 'guardrails' put in place - similar to those around human genetics and replication before we get there, rather than try to do something after the first machine unexpectedly says "Hello Dave" in an unheimlich over-calm voice. As another poster as pointed out, the people developing AI are waaay to involved in the 'how' question to pay more than just lip service to the 'should' question, that it clearly needs supra-governmental regulation. That ship has probably sailed by now though.
So on the OP's scale, if I give it more than a moment's though I'm about a 7-8 for what hell-scape we'll have stumbled into a decade from now, I don't think because of some Terminator vision of the future, but just the fact that banks, post, shopping, interactions with corporations and govts, probably some medical and light prescriptions, some music art and literature will be mostly AI, and that's as depressing as ****. It was supposed to do our drudge to free humans to be creative, but there's no money in that, so it'll end up the other way around.
I don't see a future where the machines do everything while we have 24/7 leisure time.
You're right. Mark Zuckerberg will still need someone to scrub his toilet.
Edit - just to expand on that comment a bit. Surely 24/7 leisure time is something to be aimed at? We don't have to spend the 24/7 playing Bingo on our phones or trudging round shopping centres, the way we do now (apparently). We can devote the resources to education and creativity, and sport, and generally worthwhile stuff. The alternative vision is the one where Zuckerberg and pals sit on top of a pile of everyone's money, while the population are obliged to perform menial drudgery for the privilege of staying alive.
It's difficult to see the reality from the hype at the moment.
It makes a LOT of mistakes. You have to check everything. Building autonomous agents on the back of this seems crazy to me.
You can run reasonable models locally quite easily with https://lmstudio.ai/ now for free. It makes me wonder where they'll make the billionds they've invested back again. I think all this hype is that process happening, trying to get as many companies as possible to build the infrastructure around their propietary non-replaceable tech stack, because the underlying AI model is commoditised and interchangeable.
The rich will get richer and everyone else?.............****ed
The rich will get richer and everyone else?.............****ed
That's my take on it...
People will simply be fired and competing with an ever growing pool of other people that got fired, for a shrinking pool of jobs.
Plus... Listen to this... One of the godfathers of AI being interviewed.. It's grim to put it lightly.
The 24/7 free time is a great idea, but we still need people to know and understand how things work and how to fix them. Think of the Pixar film Wall-E where the AI ship runs everything and no one actually knows how things work or how to fix anything.
The alternative vision is the one where Zuckerberg and pals sit on top of a pile of everyone's money, while the population are obliged to perform menial drudgery for the privilege of staying alive.
On the other hand there will always need to be a reasonably well heeled portion of the population to buy the goods that make the billionaires richer.
If it gets rid of tedious jobs then good. Although we had fun this week using AI summary on a report our client had overpadded using Ai. Hopefully it will reduce reports to a series of bullet points in the first place.
suspect it will be useful especially in medical fields where they could be trained to do a gp job quite easily (lots of knowledge interpretation) leaving us to check/validate
also like the fact we are already finding the fun in it. Italian brain rot is just having fun with AI slop
Just to say, long language models are the AI used to edit reports etc, that's a strand of AI, it's the Artificial General Intelligence I'm concerned about, the LLM is mostly harmless
Just to say, long language models are the AI used to edit reports etc, that's a strand of AI, it's the Artificial General Intelligence I'm concerned about, the LLM is mostly harmless
"If it gets rid of tedious jobs then good."
So those only intelligent enough to do said 'tedious jobs'., what happens to them?
Universal basic income...? Don't make me. Laugh.
They'll just be left to rot.
Yey.
Plus... Listen to this... One of the godfathers of AI being interviewed.. It's grim to put it lightly.
Set a calendar reminder for 23 months time and see how many of those jobs no longer exist
I’ve already encountered the totally unrealistic expectations of what can be done using AI. While working last year for a particularly awful company the ****-wit of an MD came into the studio, very excited.
He then proceeded to breathlessly tell us that we didn’t need to be spending so much time doing all this photoshop work and stuff as Adobe software now had AI which ‘could do a lot of it for us’.
We all looked at each other and sighed, with an ‘are you going to tell him, or shall I?’ expression on our faces. Someone took him to one side and burst his bubble by informing him that we’d all been doing just that for ages, as that was the only way we could meet his already insane deadlines.
He was crestfallen and was very disappointed that he couldn’t use this as justification to cut the timescales for jobs yet further.
He did anyway. I told him to stick his job and left at that point. So did everyone else. Good luck with replacing all the highly skilled and experienced photographers, retouchers and designers with AI, dickhead!
I’m sure similar nonsense is already pretty widespread
AI is good for removing or replacing time-consuming donkey work - and it’s still very hit and miss at that - but it can’t do the stuff that requires real skill and a degree of finesse
Define tedious too. Rather than as 'boring' if we consider repetitive and essentially the same each time, thats just the kind of things that robots do now where it used to be hands. And I think robots haven't replaced the workforce, rather enhanced it.
The example I cite is related to some work we were involved in - using ai to examine mammogram scans. I don't remember exact details but the ai bot was trained on some images and outcomes, and then let loose on another set to see how well it did. And an ai bot could get through a lifetime's worth of scans in a weekend, without getting bored, tired, etc. And not only way faster but actually was not only spotting cancer cells with the same accuracy as the human radiographer but was spotting cancer forming way earlier (in some cases years sooner)
Different study but with some data
https://www.breastcancer.org/screening-testing/artificial-intelligence
AI is delivering diagnostic tools that will save lives. That’s a whole different field to mining and scraping the combined output of human workers to suppress their wages, devalue their jobs, and further concentrate wealth within an oligopoly of USA sited companies owned by libertarian anti-union nut jobs.
My current view of LLMs… it’s like the biscuit factory in Bagpuss…
In Bagpuss the mice pretend to work a biscuit making machine… but all that happens is they put a biscuit in, and then take a biscuit out. Obviously you can’t make money this way, unless… you steal the biscuits that you put in the machine… and then use the output to your benefit.
Define tedious too. Rather than as 'boring' if we consider repetitive and essentially the same each time, thats just the kind of things that robots do now where it used to be hands. And I think robots haven't replaced the workforce, rather enhanced it.
From what I'm seeing personally, the repetitive stuff being taken away from humans, isn't really using AI but overwhelmingly soft coded automations.
Most likely negative outcome is idiocracy.
Where AI is used to replace skill, then the skills don't exist, then nobody knows how to validate the degenerating quality of AI output.
Otherwise, I'm looking forward to it taking some of the drudgery away from the routine stuff I have to do. The more interesting bits of my job are definitely not AI fodder yet.
And we'll always need people to drive the AI.
Really couldn't give a crap right now. But I do think about what this holds for my kids in the future. How my generation is the last who will know and remember a time before mobile phones, tablets, "Smart" tech. I was telling my kids how we used to have to remember phone numbers, addresses, routes to and from. If you didn't know something you'd get a book about it. There was no instant tome of knowledge for everything. Now it's "Google it".
In Bagpuss the mice pretend to work a biscuit making machine
Wasn't it chocolate biscuits made from butterbeans?
I hate it as in a majority of cases it’s going to be used to flood social media,films, music and literature and news with low quality material.
IMHO This will lead to big job loses until the public disengage with the slop and then there will be a move to people being happy to pay a premium for quality human material.
Googles ai Flow (film/tv stuff produces stunning results if you have a skilled creative operating it)
and it is going to seriously affect that area, why build sets and employee the army of people to make a tv program when it can all be done digitally, stunning results can now be achieved with way more less people.
In the medical side with the diagnostic stuff though it’s great.
The big problem with ai is it appears smarter than it is to Joe public and the tech-bros are all jumping on the bandwagon to push that.
The people who seem to be excited with it are the ones eying up how many staff they can lose.
/ I think we'll see a three-way division in society. One group, the high priests of AI - at the top. Observe the way large US companies are paying upwards of $500,000,000 to poach developers.
The second group is that which is capable of asking the AI what they want in the sense of being able to put a logical and ordered proposal into the system, and then understand the output and modify the prompt to fine-tune. I don't think the difficulty of this to get genuinely good results is really understood - it requires a very ordered and logical thought process and the ability to put this into words.
The third and final group is everyone else. I don't think that all that much will change in my lifetime. Though no doubt we'll regret outsourcing the telephone sanitisers to AI.
Ultimately I do think it'll benefit society - even if it's just as simple as letting creative people truly express themselves. A short animated film (even a year ago) took hundreds of people thousands of hours. Now it's a couple of day's work by one person. Is the visual style original? Probably not. The story can be anything that the author wants though.