AIBU to say AI will completely change how we parent and live?

lol me too .

Wow, this really makes you think about the future! It's wild to imagine a world where knowledge isn't about remembering things anymore. The idea of schools changing so much is pretty mind-blowing. Thanks for sharing these big ideas!

I know what you mean — I’ve been trying AI for little everyday things too, and it really does feel like the start of something bigger. On the one hand, it’s brilliant for quick wins (like turning a fridge full of random ingredients into a meal idea 😅), but on the other hand, it is a bit overwhelming how quickly it’s moving.
I don’t think it will ever replace the human side of parenting (the cuddles, the chaos, the instinct), but if it can take away some of the mental load, I’m all for it. Feels like it’s going to be one of those things we’ll look back on and wonder how we ever managed without.

I do think it's important not to forget the good stuff. I have been listening to AI experts on youtube this week, and some of their predictions are mind-blowing. One said that AI could halt and even reverse human ageing within ten years. According to him, by the late 2030s you'll have people in their 50s who look like 20-somethings. Another said AI could cure most human disease within ten years. Even Jeffrey Hinton, the nobel prize winner, thinks AI could cure human illness and disease within 25 years!! Right now (apparently) it takes up to ten years to develop a new drug. AI, so I'm told, could cut that to months, and then weeks, and then days. And it won't just develop drugs to cure stuff. Imagine a drug that makes you blissfully happy, or limitlessly creative, or gives you eight hours of perfect sleep, or whatever, and has zero side effects.

Thank you everyone for your interesting and informative posts.

There was a thread on here not so long ago about AI and translation, and how it would make translators and interpreters redundant.
My suggestion to confuse AI was:

A) Thank you for standing up for me last night.
B) Thanks a bunch for standing me up last night.

I guess it won’t be long before it can tell the difference - maybe it can already.

But you'd need fewer teachers, so costs are swings and roundabouts. Not bad really considering we have shortages of teachers, but not quite so bad shortages of the other kinds of workers where you don't have the same years of training/qualifications and then higher salary expectations.

@Badbadbunny I would love to spend all of my days teaching and not dealing with behaviour, nappies, toilet training, hoists, medication and first aid, but have to at the moment, as all our TAs have been made redundant.

If AI means we can afford to have me just teaching plus "social care" assistants for the "personal" care of the pupils, "assistants" to do the fetching and carrying and cleaning up after lessons etc, "security" workers for discipline and supervision rather than me doing it all, then bring it on. The budget for all those new members of staff will be vast though.

@Badbadbunny Unfortunately source and fact checking is its own “drudgery.” Right now I can do things as quickly from scratch as I can using AI - the effort just shifts to something different.

I think schools are where AI will make radical changes. A lot of those daily tasks you mention don't need to be done by a teacher. You could have "social care" assistants for the "personal" care of the pupils, "assistants" to do the fetching and carrying and cleaning up after lessons etc, "security" workers for discipline and supervision.

That means that schools would need fewer teachers, but those fewer teachers could spend all their working time actually teaching.

Let AI develop individual teaching/learning plans, produce personalised resources, worksheets, homework questions, etc., and then AI can "mark" work done and provide the teacher/parents with detailed, personalised feedback as to achievement and suggestions for improvement, but of course, it can use it's own feedback to produce new resources either to progress to more difficult topics, or if not done well, keep on the same topic but produce different ways to try to teach and garner understanding, until the pupils gets it and can move on.

Teachers can monitor progress, step in when things aren't progressing well to help understanding etc, but also step back for motivated/able pupils who are progressing well enough on their own guided by AI.

It's nonsense that teachers can spend most of their day not actually "teaching" but spending their time on discipline, personal care, social care, etc., and what little time left on planning and marking and administration.

@Dappy777 Or the big tech companies will use it to control everyone and the information they have access to, even more than they do now.

Putting so much power in the hands of a few billionaires, at the expense of just about everyone else will never end well.

I use ai for nothing right now and currently have no interest in doing so.
It’s not changing very fast in this household.

The thing we need to remember is that there's no end point. So far as we know, AI will just get cleverer and cleverer and cleverer. If we can make it benign and good, then we could create heaven on earth. If we make it malignant, we'll destroy ourselves. Simple as that. Sooner or later we'll reach the singularity, the tipping point at which AI is cleverer than the cleverest human. But it won't stop there. It will be twice as clever as any human, then four times, then ten times, then a thousand times, and so on.

The only thing we can be 100% sure of is that someone will use it for evil. Putin's first thought will not be "how can I use this new tool to rid the world of mental illness and cancer and poverty?" His first thought will be "how can I use this to dominate Europe and make Russia a superpower."

That's why we'll always need experienced professionals to sanity check the output of AI. Let it take away the drudgery, but still check the output. Just like how professionals like accountants and solicitors use junior staff to actually do the donkey work but always check the end result before signing it off. AI will take over that donkey work, but no sane professional would take AI as gospel without any sanity checking.

I quite like Wes Streeting and his approach to the NHS problems as he doesn't appear to want to shy away from big changes which is what is needed.

AI could revolutionise the NHS. Obviously in clinical matters like analysing scans, x-rays and blood test results which are currently prone to Human error.

But more significantly with the whole problem of administration and management. Let AI into the myriad of different systems and start learning how to streamline things and increase efficiency, i.e. knowing that a blood test has to be precisely x days before chemotherapy treatment, knowing which pharmacy has stocks of a particular drug in short supply, searching appointment databases in different clinics, different hospitals, different trusts, etc., to give patients a choice of quicker appointment times if they're willing to travel a bit. So many areas where "humans" slow down the process.

My OH needs numerous hospital and clinic visits, treatments etc., for his treatable but incurable cancer that he's had for around 7 years. It's a sodding nightmare as all the different departments are incapable of communicating with eachother. If, say, the oncologist refers him to orthopaedics for a bone problem, he'll just get a random appointment through the post several weeks later, which almost certainly will clash with an existing treatment or medical appointment, so he has to phone (multiple attempts to get through as they don't answer!), to cancel and arrange another appointment, etc. A monumental waste of everyone's time. Then he'll ask if there's a sooner appointment somewhere else, and you'd think he'd asked the appointments clerk to sacrifice her first born as it's "such an effort" to look at a different hospital/clinic in the same trust, you can hear the sighing and almost hear the eye-rolling! And if nothing local within a few months (bearing in mind most of his referrals are marked as urgent because of potentially relating to his cancer and chemotherapy), and he dares ask whether there are appointments quicker in our neighbouring trust across the county border - you can almost hear the appointment clerk's head imploding as that no doubt means logging into a different system to check! AI could do all that in a few seconds, produce a list of potential consultants/clinics with timescales and locations/distances and save hours of admin time for each appointment.

Same after treatments, admissions, tests, etc - it could review the latest results against previous results and all other clinicians involved and flag up contra-indications, potential drug conflicts etc., to the relevant doctors/consultants.

Finally, the potential for "joined up" thinking and actions in the NHS. Can't wait!

AI (or rather, Large Language Models like Chat GPT, which is what we’re talking about specifically) work by predicting the next word in a sentence based on existing writing/what’s gone before. They don’t fact check or verify. (Or even do maths accurately, it’s been found.)

There is a growing number of computer scientists who are very concerned at how oversold LLMs are. They generate plausible sounding text and excellent images, but they do not reason, fact check, or source check. And they make it VERY easy to generate reams and reams of LLM slop across the internet.

Like anything else new, I think ultimately there'll have to be controls etc., Trouble is, like anything new, politicians are far too slow to realise problems and take years to come up with controls etc., by which time it's fire-fighting. Such a shame that politicians are never pro-active. As there WILL be pain from AI before things get too bad that politicians will finally have to do something to impose rules and controls, etc. They have no foresight!

There's already been a tax tribunal case where a woman defending herself against HMRC challenge produced a few court case/tribunal rulings to defend her stance. Luckily, someone in the prosecution team took the time to check the tribunal/case precedents and found they didn't exist and had been fabricated. The woman obviously lost her case as she had no real statute nor precedent to defend her stance.

What wasn't made clear in the reporting was whether she was a "victim" of made up cases via AI etc and genuinely believed they were real cases, or whether she made them up herself and hoped the prosecution wouldn't notice.

That's why it's always worth performing a "sanity check" on anything generated by AI. And also why "professionals" will always be needed to supervise AI to check it over, just as they currently "supervise" trainee and junior staff to check they're doing the work properly. It's the junior levels of staff who'll be the ones at risk from AI - the experienced/senior professionals should be OK.

Of course that leads on to "how" people will ever become experienced in the future if firms take on fewer trainees and junior staff! So firms will have to keep taking on such staff to train them up to "fill the boots" of older experienced staff as they leave or retire. A few big firms got caught out like this during the Covid lockdowns as they made the decision to recruit far fewer graduates/trainees due to uncertainty as to how many businesses wouldn't survive the lockdowns. It was OK short term, but the year after, they realised that they had no "1 year trainees" to step up to do the "year 2" work, but the year 2 trainees had to step up to do year 3 work, and so on, creating a vacuum of work needing doing but no workers! So they had to hold some 2/3/4 year trainees back to do the work until the number of year 1's was back to normal in 2021 and 2022 to feed through again and recruit some more experienced staff from other firms (at higher cost) to fill the gaps at all levels.

It will definitely do some of those things.

Thankfully, our kids are already adults.

Oh shit RayonSunrise yes of course I hadn’t thought of that happening over time. That’s really scary.

I’m confused about these hallucinations though. There must be a quick and easy way for the AI to also check that the references it is giving are actually in existence as a real webpage? How is it able to just make stuff up? Obviously no guarantee that the webpage referred to is a legitimate one but I don’t understand how completely nonexistent things get referenced.

There was an article in The Register (IT news site) this week about the beginning of AI model collapse, which is when there is so much AI-churned out “data” out there that the AIs end up training & referencing mainly AI-created data (complete with hallucinations) rather than human-created data, let alone verified data. It’s a real problem.

Crap I’m going to start using cash from now on

Yesterday I asked Chat GPT to write me a short summary of research on a particularly issue for a report I was writing. It was two paragraph and included four references. I checked the sources and only one out of the four articles it used as evidence actually existed. Had I just used the text I would have put more misinformation out into the world for it to be used as evidence. It was a real wake up call on the risks of reliance on AI.