AI over the next few years

[Nutmuncher]

I’ve recently shifted my reading content from war, doom scrolling and political madness towards being positive for the next few years, mainly scientific breakthroughs and technological advances from AI. There’s so much to be excited about (aside from the inescapable dose of fear and nerves of war doom climate doom and politics) and I think it’s something MN should be talking about more.

AI is coming whether we like it or not, it’s going to bring with it a seismic shift for the world that’s going to be incredible but also bring with it a tricky societal transition that will impact us all in some way, jobs will change, industry will evolve, the human touch will become increasingly important. How easily we transition to that new world is another story, how will those who don’t use devices or aren’t technology native actually navigate a more connected world? The economic implications if entire industries go or certain careers are no longer needed could be catastrophic if not managed properly and that’s just the tip of the iceberg. But I want to focus on the positives so let’s gloss over the bumpy transition period for now 😅

The advances and changes we’re going to see in the coming years will make today look like the 80’s in a relatively short space of time. The pace of progress in companies such as Anthropic, SpaceX, OpenAI right now is breathtaking. People think picture editing or making dodgy FB posters whenever you mention AI but it’s so much more than that, we just think it’s bad for the environment and that it’s taking jobs away when actually alongside technology and robotics it’s going to revolutionise how we live dramatically.

Excited about-

Medical advances and breakthroughs, we’ve seen the impact GLP-1s brought, there’s so much more just like those coming in the next few years. Drugs are being discovered super fast, research is taking months instead of years and analysis of clinical trials is more thorough and accurate. As new technologies come available the medicines keep improving. Gene therapy and having targeted treatments based on our own genetics is an exciting area of research that’s currently happening, the understanding of our own bodies will be a major step forward. I have always been keen in longevity and wellness, areas which I’m watching like a hawk.

Education. AI isn’t going to be a hologram teacher (not yet anyway) instead it could help teachers tailor lessons to each child, minimise many of the laborious administrative tasks, help to identify learning difficulties earlier and much more. I think classrooms are going to look very different in the next 5 years.

Industry. Rather than replacing entire industries, AI will automate repetitive work, improve decision-making and help people work more efficiently. Some sectors will change more than others and productivity will increase along with efficiency in the businesses that adopt and adapt. I think industry and workplaces in general will soon be judged on how quickly they use new technology, those that do will appear relevant and capable and those that don’t will resemble an office if it were still using a typewriter today.

Anyone else excited?

Of course when most of what Ed predicts comes true, people will say 'why weren't we warned?' Thus it ever was.

And what's your time frame for all this 'supposing' and 'maybe'?

Individuals in small claims, yes absolutely. Companies taken to small claims often have it taken care of by their legal representatives though. My car dealership did. The people after us where an insurance company who did.

but either way my point is this is where ai is replacing legal fees, and it’s sort of ok because it’s not something young star lawyers cut their teeth on anyway.

It’s fairly obvious that the OP isn’t real and that responses are just being fed into chatGPT for her to come up with her responses.

There is not an element of human thinking in what the OP has to say. And none of it is remotely plausible.

Either that, or there are people out there who really are that thick.

I gave Copilot a chance last week. I was having brain-freeze over a simple calculation. Copilot not only gave me a hopelessly wrong answer: it went on to explain in detail its mathematical process, then offered to write me a calculator which also gave wrong answers.

When I queried it, Copilot blew some smoke up my arse while implying I was too thick to understand my own problem. It never came up with the correct answer.

Here's the problem, if you're interested. I'm making a 19cm bracelet that has to stretch to 26cm. I'm using elastic and non-stretch thread. The elastic can stretch to 1.5 times its relaxed length. What's the minimum length of elastic needed?

The answer is 14cm - this will give the 7cm expansion required.
Copilot told me (and 'proved') I needed 6cm but could get away with 2cm

It has not improved my expectations of living in an AI-run world.

I agree. No one can be 100pc pro anything without seeing the issues/ pitfalls. Op is not engaging in any critical analysis. Incredibly naive.

Oh absolutely! There is some amazing Machine Learning progress due to the availability of reasonably priced compute power. We could spin up a machine in 2 mins to do something we would have had to book time on a university supercomputer to do not that long ago. Other 'Machine Learning/AI' projects I've seen are just classic maths that run fast but the marketing angle is too good not to exploit. I suppose, technically, that's what most of AI is really - the first neural net was built in the 50s but hardware was too slow for it to be of much use. LLMs giving the illusion that something novel has been born is irresistible at the moment. The minute the mood changes, I guess it will all be rebranded as 'AI-free'. Given they didn't define what constitutes it's use in the first place, no one could refute the claim 😂

Incredibly naive as to what the government can do in this.

How can she simultaneously be not real and feeding responses into chat gpt?

I've had similarly dreadful replies about crochet stuff from LLMs. Hopeless with real world, physical problems.

It reminds me of the TV show QI with a lot of things. It gives a plausible/intelligent sounding answer. But if you ask it about anything you actually know a lot about, you realise the answers are often at best un-nuanced and at worst completely, misleadingly wrong

AI is utterly hopeless performing calculations itself. I've tried many times as have other people I know.

BUT, it utterly brilliant at devising codes/formulae to use in databases and spreadsheets etc for them to do the calculation.

There's a fundamental difference.

The user needs to learn by experience as to what AI can and can't do. You need to "speak it's language" so it doesn't confuse itself.

@Badbadbunny I strongly suspect the users finding LLMs' formulae utterly brilliant are incapable of assessing their validity.

I let Copilot walk me through its process using Excel. I let it write a calculator for use in Excel. I let it write a replacement after I queried it. All wrong.

The programming world's frantically trying to unpick labyrinthine AI-written code which, it turns out, has created multiple backdoors to critical structures and, in some cases, has protected those backdoors from being closed. People are realising their servers are being filled with terabytes of code that nobody can understand, which could and should have been written in several megabytes.

I agree the genie's out of the bottle and will change our world in unexpected ways. I'm fairly certain these ways will be fucked-up, irrational and dysfunctional.

I have Gemini at work and I've used it to build a program for all my sales analysis and forecasting. it has more of less replaced a data analyst role

"Job roles we can't imagine yet" is doing a lot of heavy lifting here...UBI for everyone is the only way it can work.

This is where experience comes it. Someone who doesn't understand what they're asking it to do probably won't notice/realise when it's cocked up. It's why we'll always still need "professionals" to supervise AI output. It's the bottom end of the working scale that will disappear, i.e. the admin staff, the trainees and juniors. We still need the "top end" to ask AI the right questions and review their answers.

Of course, that creates the problem of where the next generation of "experts" comes from if firms stop taking on junior/trainee staff as there's then no one coming through, gaining experience, ready to take over when the "experts" retire.

We saw the same problem during covid when the likes of accountants/auditors didn't take on the usual quote of graduates/trainees because they'd seen an immediate slow down in work due to client firms going bankrupt and not knowing how many other clients would go bankrupt. It meant the usual "Moving up" through the firm couldn't happen and it's taken several years to rectify and get the system working again. Some staff had to be promoted beyond their ability to fill gaps higher up and some staff had to be kept at the bottom to do the "drudgery" work meaning their progression was slowed down.

My DS who works for one the the UK's biggest insurance/pension firms says that they're actually increasing recruitment for actuarial trainees/graduates as they think there's still going to be high demand even with AI, but they're reducing recruitment in other areas such as marketing, admin, accounting, etc. They know it's a big risk, but at least they're trying to "foresee the future" and there's already a drastic shortage of qualified actuaries so it still makes sense even if AI has a similar effect on actuarial jobs as it's forecast to have on other professions.

@Badbadbunny, good to hear but I hope they aren't using AI to train the experts!

But as others have said. Two huge problems.

Where does the money come from to pay it? Taxes on those working will have to increase massively to make up for loss of tax revenue for those not earning AND paying them benefits. No one is going to work if income tax etc rises to stupidly high levels to bridge the gap.

And more importantly, how does society cope with huge numbers of bored people with little/nothing to do, the mental health problems of people with no aim in life - we already know the societal problems in areas of high unemployment often result in poor health outcomes, increased crime, increased substance abuse etc. We'd need to find things for people to do to occupy their time, occupy their minds, give them a sense of purpose etc.

I just did a big code release and every single line was written by AI. It's so much better to use computer written code than writing by hand. Saves loads of time and reduces errors.

Ha!

I'm the poor smuck that has to sit for hours fixing what AI breaks.

It is a glorified text predictor. It has been fed every piece of art in the world, but can only spew out generic and immediately identifiable crap. It's 3D model generation lools great, until you look under the hood at the mess and attempt to scale that up for a full production. And it's confidence at being wrong when coding is worse than the newest intern.

Add to that the sheer unprofitability of every AI company, data centers that will be obsolete before they can ever be used, and the increasing hostility to the resources being used for little reward.

I'm not worried at all.

Though, admittedly I am worried that dumb ceo's are firing graduates, because where is the next generation who will know how to fix what breaks?

I think AI will do many wonderful things in the next 10 years and will make the a remarkable point in history similar to the Industrial Revolution and the age of the Internet. However I do have some reservations about it being used around social media in particular. We already have algorithms and reports of far right parties using it to create content that pushes those ideals at election time. Already people are using it to create cartoon images from photographs, where is the original image being used and how is that improving the technology? I think people are quite naive about this technology and aren't applying critical thinking or common sense when using it.

What frightens and fascinates me is the zealotry coming from some who rave about AI and its potential. The flavour of "resistance is futile" and "adapt or die". Bonkers.

Beyond the potential environmental and economic upheaval, we have been tempted into the biggest social and psychological experiment since the internet came into being, without informed consent because even the pioneers of this tech don't know its full capabilities. (And are backing down and out of the industry at a suspicious rate). Anything we're being allowed to "play" with is probably years behind what the military industrial complex is honing behind clised doors.

It is the ultimate propaganda tool, a tool of control, and may well render way too many people obsolete for governments to deal with.

AI psychosis is a new and real phenomenon, murders have happened because of ChatGPT type models feeding into a persons deteriorating mental health.

It is nothing like previous technological advances in terms of speed and scope of change and upheaval.

Given the parlous state of the world in terms of financial, political and tribal division and instability, I am highly doubtful that the tech bros, some of whom have disturbing underlying ideologies, are going to be falling over themselves to keep the masses happy or fed if they don't need them.

And one very overlooked and minimised issue is what happens when our totally tech infrastructure goes down, by natural or man made means?

Honestly, the AI Utopia is as much a hallucination as the AI slop dominating every facet of our lives now. We have engineered our own obsolescence to a large degree, and it really isn't going to end well.

Did you use AI to write this post?

@Badbadbunny

You raise good points re people finding purpose and meaning but wrt affording it. In my head the "money" to pay for it would be a sort of World Bank controlled "Monopoly Money" that each country received while a "real" economy still took place alongside it. That's my attempt at making it make sense, anyway 😊.

Perhaps credits rather than money. The nhs are suggesting walking for 30 mins for shopping vouchers. It’s not a big stretch.

Anyway it’s not a world that seems appealing to me.

AI psychosis is just terrifying