Can someone explain AI and why it’s set to change the world?

[Cheekyfuckerry]

So sorry to ask what is probably a ridiculous question but can someone explain what it is? How can it replace humans? What is chatGPT? (I may not have that phrase correct by the way)

I keep seeing threads about how our lives will change, how it will take over all our jobs etc, I can understand that maybe it will be able to answer phones and queries like the way bots on live chat / phones do now but what else is it able to do?

How can it change the world?

Great post. In a nutshell - we're creating a system to think for itself, solve problems and come up with solutions. But the trouble is we already KNOW humanity is the problem in a lot of the major world issues - environment, war, over population etc. It's inevitable that the AI will come to the conclusion humanity must be either removed or controlled because it is correct to come to this conclusion. We could do it for ourselves, without the literal destruction of the species...but we won't.

Is this the Singularity? How will Quantum computing change things?

I'm a scientist (biochemist) in my 50s. When I started in science it was a very manual job. We made up solutions, we transferred small amounts of liquid from tiny container to tiny container, we used computers very little and those slightly older than me would tell stories about using printouts, cutting out a visual image and weighing it to determine the area under the curve. We wrote up our experiments in labbooks, did calculations by typing in data into excel, printing out tables and graphs and sticking them into our labbooks. Our labs and offices had lots of storage space for paperwork and we had big libraries full of back issues of scientific journals.

Now, we buy in concentrated stock solutions, we use robots for repetitive pipetting, all initial analysis is done using specialist software and we use statistical software to make sure the patterns we see are real (human brains tend to overinterpret data and see patterns and differences that aren't actually statistically significant). We use electronic lab notebooks to record our data and look up papers on google scholar. I now work in a virtually paperless office. The last 30 years has seen an incredible increase in automation for both physical and mental tasks in the lab. That does not mean we employ fewer scientists, it means we can do more experiments and so generate better data and so come to more robust conclusions. It also allows time to think more creatively about how to do what we can't do yet. Your list is all the routine stuff, not the crucial inventor part of being a scientist 'method A can tell us X and method B can tell us Y but if I Heath Robinson a brand new piece of kit then I'll be able to do brand new method C to tell us Z'.

There are always leaps forward with new technology but never in the way people predict and they have never resulted in overall fewer jobs. We've been predicting that technology will reduce our work for millennia but we still do more and more because e.g. freeing us up from carrying buckets of water from a well just allowed us to e.g. spend more time developing a more effective way to spin wool which freed up time to e.g. observe plants and preserve seed from the best varieties to improve yields which reduced farming time which freed up time to e.g. improve methods to make better pottery vessels which freed up time to e.g. develop a system of writing etc etc etc.

Stuart Russell on the Reith Lectures last year https://www.bbc.co.uk/programmes/m001216k/episodes/player

He explained AI as a bit like predictive text, at first it gets lots wrong, but as you interact with it more it becomes increasingly accurate, to the point where it can then 'suggest' words creatively.

AI is still getting lots wrong, but is taking off exponentially as it is being used in so many fields. Once it gets 'creative' we don't now where that will end up.

It's speed and internal interactions mean that we can't follow or predict how it will develop.

It's the most irresponsible and truly scary development. But we are all fucked anyway because we won't change how we live and it's destroying the environment that we need to survive.

Our poor children and grandchildren. What a world they are inheriting from us. 😓

It also allows time to think more creatively about how to do what we can't do yet.

But what if what you think of as 'thinking creatively' is actually a complex-but-predictive series of thoughts, and it turns out that AI is better at it than you are? What is creativity, really?

I think that’s what concerns me most. What if everything that our human brains are capable of can - eventually - be replicated by AI? If our brains are living computers, and every thought process can be replicated or represented by an algorithm, then an AI will -eventually - be at least as good, if not better (cheaper, more efficient) at it as we are. I’m not religious, I don’t believe in souls or other magic ‘essences’. But if that doesn’t exist, what differentiates us from very clever AI?

Show me an AI robot that can make a cake. That's pretty basic creativity for a human.

How do you think Mr Kipling makes cake? Doesn’t even need AI. There’s a recipe to follow, all he needs are machines that can do the physical actions required and follow instructions. There’s nothing creative about making a cake.

Agreed but how could it ever taste a cake to understand what flavours go together? Sure, it can follow recipes but could it come up with new ones. It doesn't have taste buds. Can human taste buds be replicated? Obviously, sweet, sour, bitter etc can all be replicated but in such a sophisticated way as human taste buds? They've been described as sentient (it was a very compelling and thought provoking description) but how could they ever taste a cake (to continue with the cake theme) and go, mmmmm, delicious?!

Of course AI ca come up with a new cake recipe

I see Sky News have been reading MN again 😆

https://news.sky.com/story/terminator-and-other-sci-fi-films-blamed-for-publics-concerns-about-ai-12895427

Well people need to stop demanding to wfh full time for a start. All we have is being human and interacting with other humans, if we start to diminish that it will take over so much faster.

I think it sounds like a complete and utter nightmare. I've just seen an item on the TV news about a creative event where everyone interviewed was against it and wanted people to be the ones doing the creating, not a machine. We've already lost a lot of skills due to technology, this just seems a big step too far.

@Itisyourturntowashthebath the AI didn't make the cake, it had to use a human to do that. My point is the interaction of the AI with the real physical world is not the same as a humans. Those people asked it to invent a mashup recipe but they asked it to put together flavours that they knew would work together. If you were to ask it to come up with a meat trifle mashup recipe it could do it but it would not taste good.

Using AI to come up with new flavour combinations is old news. The early iterations were a bit more Hester Blumenthal than they were Mr Kipling, but they've been refined since.

AI doesn't need to be able to taste to come up with flavour combinations. AI isn't about a thinking machine, it's about a machine that generates patterns that it has learned are 'correct'.

@JaninaDuszejko of course AI can't bake a cake in a domestic kitchen.
But it could bake thousands of cakes in an automated factory.
https://www.italianfoodtech.com/food-industry-the-role-of-automation

AI can read every recipe ever put on the internet, read every review, the task would be a piece of cake.

As for the meat trifle - search internet for definition of trifle, find Rachael's meat trifle recipe, read the bad reviews, find loads of recipes for things like biscuits and gravy, moussaka etc - invent tasty meat trifle.

AI can take every creative idea that you, I, Heston Blumenthal and another 2 billion other people have ever had and run with it.

Creativity will just be lost if AI took over. It will just be logic and programming? There won’t be a need for creativity in a human sense

As someone who worked in the CJS for many years, I’m wondering if it could/would take over from Judges and juries? Theoretically it could probably analyse all the evidence and come up with a guilty or not guilty verdict. There’d be no bias at play. Sentencing would probably actually be fairer as it would be consistent across the board. No need then for legal representation if the verdict is just analysed/decided by AI.

Anyone managed to sign up to chat GPT? It keeps rejecting sign up as they’re at capacity.

Several of us. At least three of us have posted output from ChatGPT to this thread!

It does take a few days for your invite to use ChatGPT to come through after you request it.

Just keep an eye out for companies called Skynet, and extremely good looking muscular men who appear naked in the middle of the road unexpectedly, surrounded by lightning 😀.

I want to know this too!

@EmpressSoleil that’s coming, for all the reasons you outline. Human judges are demonstrably inconsistent in sentencing - depending on whether they’ve had lunch or not, among other things.

I was listening to Matt Frei on LBC this morning - he had an expert on who said that human annihilation wasn't "inevitable or imminent". Matt asked if that was the best comfort he could give...