A.I. (general thread for any AI-related topics)

The information used by the Natural World is real, it is true, verifiable and reproducible, there is no debating if something is truth or not. The information is universal, not regional. We can view it by using math, chemistry, and physics.
You see, I don't agree with that statement. There are regional "truths" - which I pointed out earlier with black and white swans, and there is debating of the "truth" - you only need to watch any goal scored during any Football match (even with VAR, different fans will see it differently, the VAR was viewed wrong, the referee was blind, rules weren't applied, rules were wrongly applied.)

And I think that is where it begins. Even humans cannot agree but the machine can only accept yes or no. @Wayne Mack said machines just need better training. You mentioned training AIs in their "natural environment" but what is more natural than Maths and yet they still get the Roche Limit question wrong!
 

Why A.I. Isn’t Going to Make Art​

To create a novel or a painting, an artist makes choices that are fundamentally alien to artificial intelligence.


Sorry for the paywall.
But the gist is that since AI doesn't understand language, it just imitates word order, it cannot create the new ideas that make for a reasonably good or innovative novel or painting.
 
The organization that runs National Novel Writing Month says condemning people who use artificial intelligence to draft books is "classist and ableist."

 
You see, I don't agree with that statement.
As soon as the word truth is used, there has to be a debate. I was thinking if you got through enough layers of the onion, you would arrive at information that truly represents what is there. Humans are not operating at this level. The debate about what something is, or is it truth, is based on our interpretations of what we think the universe is made of. I should have said at the most basic level, it is verifiable and reproducible but we can't do it with the tools we have at this time.
 
The organization that runs National Novel Writing Month says condemning people who use artificial intelligence to draft books is "classist and ableist."

Wait until Copilot adds the 50,000 word option to its standard output of Paragraph, Email, Ideas, Blog post with a length of Short, Medium, Long.
 

Why A.I. Isn’t Going to Make Art​

To create a novel or a painting, an artist makes choices that are fundamentally alien to artificial intelligence.


Sorry for the paywall.
But the gist is that since AI doesn't understand language, it just imitates word order, it cannot create the new ideas that make for a reasonably good or innovative novel or painting.
I think there's a huge mistake going on now with Large Language Models continuously being called AI. It's all just marketing hype. Which results in questionable statements like this saying that AI, meaning things like ChatGPT which isn't really AI, can't do x, y and z. There is no real intelligence in LLMs just a vast database of word probability orders and so to call them AI is fundamentally flawed.

This is, I think, a great danger because these sort of nascent technologies are being judged against what a human mind can do which is about as sensible as expecting an 18-month-old baby to come up with the theory of relativity. It's not going to happen and consequently 'AI' is condemned as not being intelligent.

I find it strange how such a young technology keeps getting condemned based upon is current performance against human minds, what do people expect? This is a technology that is coming and will probably improve exponentially over the next decade or two. Simply listing all the things it can't do now is like early cars being condemned because they couldn't go as fast as steam trains. It's silly. I don't want to know what AI can't do today, I want to know what it can do. And yet every time there is some new advance in what the emerging AI systems can do, that news gets drowned by a barrage of accusations about what it can't do.

I'm sure that no one who has children would have expected them to come home from primary school expounding the origins or life, the universe and everything.
 
I can see a threat of AI replacing visual artists (immorally via reuse of their past works), albeit with some human at least capable of making corrections, but for writing... aside from matters of translations, I can't see it working out practically. If somebody wants a program that will be a deluxe spell check for grammar and other things, that is going to be more than a little problematic... requiring of the kind of skills of a human corrector or clean up person afterward as to really make using such a program or AI superfluous!

Beyond that what human is going to be willing or at all satisfied to read or otherwise experience the 'creativity' of a simulation other than at a novelty curiosity level? We want read with the expectation of experiencing what another person thinks, creates, has to say for themself. from a poem or lyrics to a thousands of pages work of fiction. How much time never minding money would you invest in anything like a processed simulation. Granted, on the surface the AI 'product' might be indistinguishable from a great many human authors (and as with visual art it would be helpless without past works of humans fed into it), but the chance of something extraordinary is below even random chance of the billion monkeys, and isn't that possibility some part of what keeps an audience there and seeking?

I won't deny a machine might, drawing upon past works, come up with something which could impress, but I think that would at best still be a product of an ignorance in the beholder of what is being drawn upon, and also quite possibly an accident, not proof of a human level imagination. If you factor out the real imagination unique to the best of humans I think you have a lot of input totaling zero ultimately.

As far as I'm concerned they can do whatever they want or can with AI in terms of composing text, it will never succeed outside an experimental exercise with only the programming always being the actual 'art'. Language is simply something too inherently human and listening to or speaking with a machine is always something else if not less for us. As much a waste of time basically as asking an inanimate appliance to form an opinion (at least outside of gestalt therapy).

AI produced visuals have a human direction via speech and/or text... that could become some sort of a skill as to what to input, but with text, if you are telling a program to write a romantic cowboy story with flying stagecoaches and sentient cacti... well, you might as well just go ahead and enjoy writing it yourself to get the story you want and how/why you would tell it as to just feed some basic skeleton in. I think even the dimmest bulbs of Hollywood will calculate the odds against them very long, assuming of course there is a real audience that can make something a success or determine what's a dud. If a writer wants to play with AI and can get something useful to them from it however that seems fine to me, but asking an audience to take time for or to buy what would be at best an early draft or some rough material for possible use in something more finished... it would have to be a very poor audience very easily entertained or satisfied. I realize that describes many things on television now which somebody somewhere presumably does watch, but that kind of content is of little value really and in ample enough supply.

Money that doesn't at least imagine hitting a jackpot, which an AI creation could never really be capable of, can go toward pork futures or some other commodity and have much better odds. They can have 100% of a horse that can only come in fourth or fifth, and that will be what will be shown over time I believe. If I am proven wrong it's fine with me, but it will take time to find that out, and I have no interest in condemning... go right ahead. After it reaches a limit with the bugs worked out we will still end up straight back with our fellow humans who, at least the rare few we often label geniuses, are without limits in some key regards, that can take 2+2 and find it equaling something other than 4.
 
I wrote this, then submitted it to copilot for a funny rewrite, paragraph format, medium length. I figured funny would allow acceptable mistakes to be made.

There were two problems, the first was that it works in 2 K batches. I had to split the original article into 2 pieces. So there is, and there isn't continuity between the 2 2K pieces, because it does have the previously entered information in it's memory to which it does refer when figuring out what words to use. The first piece came out immediately, no problems, it had paragraphs in it. I entered the second part, and it lost the prompts twice or so it said. The prompts were there. I had picked humorous from Professional, Casual, Enthusiastic, Informational, Funny. Picked paragraph from Paragraph, Email, Ideas, Blog post and then Paragraph from Paragraph, Email, Ideas, Blog post. The length I picked was medium, from short, medium, long. Nothing came out. Tried it again, nothing came out. Next attempt it blurted out a couple of lines of letters, looked like words, sort of, but no real words, some without vowels. Tried again, this time it spit out the rewrite but without any paragraphs.

My version:
Cultivated meat, formerly known as lab grown meat, has gone way past science fiction fantasies. It's also gotten to the point where people are trying to outlaw it. Some are trying to make it a felony to sell cultivated meat to the public.
The cultivated meat industry was hoping to cash in on the perceived waning of enthusiasm for the vegetable based substitute meat products. The race is on to find cells that reproduce rapidly so that the meat can be churned out like sausage.
Italy and a few US states, Florida, Alabama and Iowa, have passed laws banning the sale of cultivated meat. France is working on legislation to ban cultivated meat.

The European Union has banned the cloning of farm animals that would be used for commercial purposes. The US has not banned cloning of farm animals for commercial purposes.

Cultivated meat can be grown in vats. It starts as a few cells which are fed nutrients from which grow into lumps of meat. Companies have been working on scaling up the operation to make it cheaper and easier to make.

There are two basic ways to grow the cells to a sizable, edible pile. One method is to create a structure for the meat cells to anchor to, those designs are subject to a great deal of imagination. Artificial cellulose has been tried to mimic a tissue environment. Chitin is a natural scaffolding material for cells to grow on but enzymes can break it down, requiring careful planning. Collagen is another natural tissue growth anchor, but controlled production of the collagen is a bit of a problem, one solution is to get it from tobacco plants.

The roots of mushrooms have been used to create meatless bacon that looks pretty much like a piece of well cooked bacon. One company is using a dietary fiber called Nata de Coco, derived from coconuts, to create nanocellulose sponges for their scaffolding.
Another method is the bioreactor which are closed cylinders that use mechanical, chemical, and energy sources to produce the meat. It might be capable of being portable or even installed in residential settings. When there is enough meat produced, it has been proposed that the meat could be formed into usable shapes by 3D printers.

Around 40 companies are working on cultivated meat, a few are producing it in commercial quantities. Some are working towards being stand alone companies while others will be absorbed by big corporations that are looking for ready made solutions.
Big name meat production companies such as JBS, Tyson and others are actively pursuing efforts to produce cultivated meat for large scale use. For them it's a source of meat that is not subject to the uncontrollable effects of climate change, a product they can produce themselves.

Cultivated chicken is for sale in Singapore, but only 3 percent grown chicken cells. The rest is protein and spices. Right now it is too expensive to sell pure cultivated meat to the average consumer.

US, China, Israel, Australia, Singapore, Canada, Hong Kong, Czech Republic, Turkey, Spain, India, Switzerland, United Kingdom, Japan, Netherlands, are actively pursuing cultivated meat programs with a goal of providing meat to everyone from the rich to those with not enough food to eat.

A parallel industry is using bacterial fermentation to produce livestock by products. This includes milk, honey, eggs, cheese, gelatin, chitin, dairy proteins, egg proteins, sweeteners, enzymes, vitamins, fats, and pigments. All of which is being used commercially, though some are in limited quantities.


The copilot rewrite:
In the not-so-distant future, the meat aisle could look a bit like a sci-fi novel come to life. Imagine this: instead of picking up a pack of chicken or beef, you're selecting your preferred brand of cultivated meat – grown not on farms, but in vats, where cells dance the cha-cha as they multiply into your next dinner. It's a world where "farm to table" takes on a whole new meaning, and the only "cows" involved are the ones in virtual reality headsets, blissfully unaware of their digital pastures.

But wait, there's a plot twist! Some countries and states have decided that this vat-grown venture is not for them, passing laws faster than you can say "cellular agriculture" to ban the sale of this futuristic feast. Meanwhile, the industry is racing to find the Usain Bolt of cells – ones that reproduce at breakneck speed, turning vats into veritable meat-making machines. The goal? To churn out cultivated meat like it's going out of style (or before it's outlawed, whichever comes first).

Across the pond, the European Union has given a firm "no" to cloning farm animals for commercial munching, while the US is still mulling it over, perhaps waiting to see if cloned cows come with a side of superpowers. And as for scaffolding? Forget steel and concrete – we're talking chitin, collagen, and even mushroom roots. Who knew fungi could be the foundation of your next faux-filet mignon?
So, as the cultivated meat saga unfolds, one thing's for sure: the future of food is looking meatier, and a whole lot more interesting. Just don't be surprised if your next burger comes with a backstory of cellular heroism and legal drama – it's all part of the charm of dining in the 21st century.

You might just find yourself saying, "Honey, the steak is printing!" as bioreactors hum in the background, churning out tomorrow's dinner. Imagine a world where your kitchen counter has its own meat-making machine, right next to the toaster and coffee maker. It's like having a tiny, unassuming farm that fits between your blender and spice rack, minus the cows and the clucking chickens.

And let's not forget the companies in the race to become the 'Apple' of cultivated meat, each vying to be the first to get their 'iMeat' into homes. Meanwhile, big meat corporations are turning into high-tech farmers, swapping their cowboy hats for lab coats as they wrangle cells instead of cattle. As for the global pursuit of meat without the moo, it's a veritable United Nations of protein production, with countries banding together in the quest to feed the world, one petri dish at a time.

And for those with a taste for the exotic, there's always the chance to sprinkle some lab-grown honey on your bio-fermented cheese, because why not? In this brave new world, the phrase 'farm to table' takes on a whole new meaning—it's more like 'lab to table,' with a pit stop at the 3D printer. Bon appétit, future foodies!
 
I think one difference between traditional software and AI software is that all the presets are set before the traditional software runs. AI software starts without all the presets set, the unset presets get set by the program as it runs. It seems like the AI software is making decisions but it is only selecting from a list of preselected choices, it can't make anything up.
 
AI chatbots might be better at swaying conspiracy theorists than humans. Interesting article, definitely looks like one size doesn't fit all.

Deprograming works best one on one, one person to another. Its probably not farfetched but the same program could probably multitask several patients at the same time, So much for the one one one limitation.

The program works good because its the same as a person interacting one on one with a conspiracy believer, communicating with them on a personal level, about what they are thinking, which one of the most successful ways to deprogram someone. Its also one of the best ways to get other people to see your point of view, maybe even appreciate it.

The deprograming program is heavily medicated, that is, it's personally programed and filled with only handpicked facts so it can't go off the rails. Sounds like psychoanalysts should be concerned about their job futures. I'm thinking computers are going to turn the medical industry upside down. From auto workers to psychoanalysts, where are they going next.

The reported results were that people interacting with debunkbot had their confidence in conspiracy beliefs reduced by 20 percent. They compared those results against other kinds of interventions which yielded 1 to 6 percent. Turns out the other kinds of interventions were minimally trained AI programs for the job that needed to be done. The article did mention that people who can easily get “heated and disrespectful,” or friends or family members judging them for altering their beliefs, would be less successful, and forgot about professionally trained people who have a much higher success rate.

One researcher thought people would be unlikely to interact with debunkbot on their own. He suggested people could also use debunkbot to quickly and thoroughly fact-check new claims they’ve heard, cultivating a healthy level of skepticism and making it less likely that they will fall down the misinformation rabbit hole in the future, Costello notes. “You can almost think of these chatbots as a form of epistemic hygiene,” he explains, “like brushing your teeth, but for your mind.”

This is a good use for AI programs, fixing the damage done, but it also illuminates the path for AI programs to be even more convincing when misleading people. Will debunkbots be trained to overlook conspiracies that exposing instead of curing.
 
I find it strange how such a young technology keeps getting condemned based upon is current performance against human minds

It's not strange at all. Until they discover otherwise, people have a tendency to believe that something does what it says on the box.

The problem is that too many creators and/or marketers of new things seem addicted to (often grossly) over-promising and mislabeling. They are to blame for calling something intelligent** when it is not intelligent at all.

One cannot blame the person in the street for noticing the promise in the name is not being met... or, alternatively, for sometimes believing the often nonsense output of so-called AIs.


** - Calling it artificial intelligence is not really a clue to its lack of intelligence.
 
I uploaded my time travel short story, Clocking Time, into Google's experimental notebooklm.
The seven minutes podcast was one of the outputs generated by this tool.


The podcast had a few minor grammatical errors but I felt the two AI-voices captured the essence of the story.
 
Screenshot_20241007-011438_Gallery.jpg

Made in AI

Please feel free to use for a writing prompt.
 
More or less - I asked Gemini to prove it and it gave me something like this:

STRAWBERRY:
S=1
T=1
R=2
A=1
W-1
B=1
E=1
Y=1
There are 2 Rs in STRAWBERRY. Thank you for checking.
 

Similar threads


Back
Top