@Kyeh There was a guy recently(that didn’t last long, but for other reasons) that seemed like he was only semi literate. The job requires very little reading and writing but things needed to get signed off and dated. Somehow that was a challenge. Super basic math was also a bit of a struggle. He was perfectly proficient and playing on his phone though. Your fear may already be coming to fruition.
@metaphore Plus remember all the hysteria over “Y2K?” Only now it would be even more crippling, wouldn’t it? And it just seems inevitable that one of these days that’s going to be a massive solar flare, or an act of sabotage, and all the connections we take for granted now will vaporize …
One of the favorite quotes I heard was “help us solve Global Warming” and the A.I. answer would be “simple… eliminate the humans. Let me take care of that for you.”
@pmarin essentially… Machines use less “resources” (excrement, maintenance, rest) than humans as measured year to year for the average lifespan. They can also survive in more extreme conditions, like say on a planet with a carbon-rich atmosphere. Human engineering hasn’t gotten to the point where we can modify ourselves to “catch up” with the ecological change, and we can’t modify the environment to stay the same.
We take AI’s inaccuracies as truths. AI is creeping into doctor visit notes. A scary thought. My PCP has been in two research studies trying it out and says it makes too many important mistakes. Other research found it is more or less fine on a simple sore throat but the minute something gets complicated or there are multiple issues it screws up. I’d guess there are some doctors who’d never go in to fix the mess. My doctor said in both studies it took more work to fix the mess than if he had written the notes to begin with himself.
@Kidsandliz@narfcake@pmarin
Reminds me of the interstellar humans in WALL-E where as a species we’re so used to being smothered and provided for, we become massive lumps of fat
@Kidsandliz true.
I was using AI to assist with a literature review for work. One piece of data it gave me I couldn’t find anywhere. I did several searches before asking AI where it came from.
It replied that it couldn’t find any sources to match the data.
I worry about self driving cars using googles captchas to train them when you have to click both motorcycles and bicycles to satisfy it (I’ve hit this one multiple times). Or busses and 18 wheelers, or palm trees and mountains. All are real examples of what I have had in the past.
Remember Amazon touting about using AI for their “Just Walk Out” at their markets? Turns out that AI didn’t really mean Artificial Intelligence; instead, it meant Amazon India.
AI models learn from data. If this data reflects existing societal biases or underrepresents certain racial groups, the model will inherit those biases.
I have no mouth and I must scream.
Written by Harlan Ellison, published in 1967.
There was a supercomputer AI called Artificial Mind that was the result of all AI on the planet merging. And it hated us. And it reduced the Human race down to five people that it kept around to torment.
So there’s that.
@PocketBrain yeah people don’t realize A.I. is not a new thing, just the super-powerful chips and huge database of all of us called the internetz, which didn’t in those days.
Might also look up “the Adolescence of P1” I think it was.
@PocketBrain One of my favorite book scenes is in Asimov’s Foundation. The Imperial Library on Trantor had “reader stations” where book/document texts could be read on a screen. If a reader were to rest his eyes on a single word or phrase for a couple of seconds, it would “jump” to a definition of the word or a longer explanation of the topic. So: hypertext, as described by Asimov, circa 1950.
@pakopako@PocketBrain eek, you are right, but that is probably inevitable. Our education system has probably slipped so much that almost none, especially the politicians, who cut class to go smoke behind the football field, have no idea what an LCD is, and I don’t mean your screen.
You want to spur on those who can figure out a way to create something for both you and me, kind of okay with those that believe in self-sacrifice so at least someone gets theirs, be mindful of those who will betray you to suit their own needs, and avoid the unpredictably of a stupid person at all costs.
@pakopako@PocketBrain Long before semiconductor chips, this was shown by the Greeks and then Romans. amazing geniuses and inventors. Political corruption. Occasional assassination “et Tu, Brute” allegedly. So you are saying the A.I. will both advance and degrade our civilization at a much faster rate? Oh the upcoming “dark ages” will be fun.
@ebatch Elon is up to about 10 launches a week now, each with 20-25 starbots. I think when Jeff Bezos launched his first batch for his own network, Elon ramped up his launches. Always wonder if the starbots will have space-to-space combat and the truth is we would have no idea what is happening up there.
There’s no AI yet, Large Language Models are just glorified autocomplete, and useless without large scale theft of intellectual property for training.
Currently I think the biggest problem is they lie with same certainty that they have when getting it right. What use are they really, when you have to double check their output? And since this is a well known problem, why don’t the LLM makers have a verification step built in?
@blaineg@pmarin I find it “interesting” that the “are you a human” test grids are AI-generated and checked. So we have to convince a computer that we are human - the opposite of passing a Turing test. I don’t think Alan Turing ever saw that coming.
On our topic, look up the movie Dark Star. I didn’t realize until recently it was directed by John Carpenter. It is kind of a hippie-druggie-sci-fi-comedy. (It was the 80’s)
Aside from entertaining combat with a giant alien beach ball, the main point was the “smart bomb,” which was told its purpose was to explode. But it was a glitch, so they tried to stop it (don’t want to spoil more than that). They (commander and the smart bomb) eventually discuss “phenomenology,” and 40 years later I’m still not sure what that is.
@pmarin@Kyeh I remember watching Dark Star (in the 70’s) at the midnight movie at Fifth Avenue Cinema in Portland (high, of course). It’s available online (free, multiple sources) - my wife and I watched it recently - still fun, but not as much as I recalled from my original viewing. Pretty good/fun SFX, given the shoestring budget.
All of the above, plus the environmental costs. And humans getting stupider because they can just have AI do it.
@Kyeh There was a guy recently(that didn’t last long, but for other reasons) that seemed like he was only semi literate. The job requires very little reading and writing but things needed to get signed off and dated. Somehow that was a challenge. Super basic math was also a bit of a struggle. He was perfectly proficient and playing on his phone though. Your fear may already be coming to fruition.
@metaphore Plus remember all the hysteria over “Y2K?” Only now it would be even more crippling, wouldn’t it? And it just seems inevitable that one of these days that’s going to be a massive solar flare, or an act of sabotage, and all the connections we take for granted now will vaporize …
One of the favorite quotes I heard was “help us solve Global Warming” and the A.I. answer would be “simple… eliminate the humans. Let me take care of that for you.”
@pmarin essentially… Machines use less “resources” (excrement, maintenance, rest) than humans as measured year to year for the average lifespan. They can also survive in more extreme conditions, like say on a planet with a carbon-rich atmosphere. Human engineering hasn’t gotten to the point where we can modify ourselves to “catch up” with the ecological change, and we can’t modify the environment to stay the same.
We take AI’s inaccuracies as truths. AI is creeping into doctor visit notes. A scary thought. My PCP has been in two research studies trying it out and says it makes too many important mistakes. Other research found it is more or less fine on a simple sore throat but the minute something gets complicated or there are multiple issues it screws up. I’d guess there are some doctors who’d never go in to fix the mess. My doctor said in both studies it took more work to fix the mess than if he had written the notes to begin with himself.
@Kidsandliz Now what’s wrong with eating rocks everyday? It’s
geologistdentist recommended!@Kidsandliz @narfcake anybody remember:
If you put fruity pebbles in your mouth, you will never have rocks in your head.
@Kidsandliz @narfcake @pmarin
Reminds me of the interstellar humans in WALL-E where as a species we’re so used to being smothered and provided for, we become massive lumps of fat
@Kidsandliz true.
I was using AI to assist with a literature review for work. One piece of data it gave me I couldn’t find anywhere. I did several searches before asking AI where it came from.
It replied that it couldn’t find any sources to match the data.
@Kidsandliz @kittykat9180
So it made it up?
@Kidsandliz @Kyeh it seems so.
Neither the I nor AI could find a source to support it.
https://shirt.woot.com/offers/natural-stupidity
@narfcake
https://www.buytshirtdesigns.net/t-shirt-design/its-fine-rpg-gamer-cat-d20-dice-fail-funny-nerdy-geek-t-shirt/
(I can’t find the “well you rolled Critical Stupid” shirt I saw last week)
I worry about self driving cars using googles captchas to train them when you have to click both motorcycles and bicycles to satisfy it (I’ve hit this one multiple times). Or busses and 18 wheelers, or palm trees and mountains. All are real examples of what I have had in the past.
Beep beep boop boop. (I assume if you failed CAPTCHA you now understand Robot.)
@Kidsandliz
https://xkcd.com/1897/
@Kidsandliz “choose all the armed civilian combatants in the photo”
@jouest Wait. What? These self driving cars have phasers to take out the armed civilian combatants? Cool.
Remember Amazon touting about using AI for their “Just Walk Out” at their markets? Turns out that AI didn’t really mean Artificial Intelligence; instead, it meant Amazon India.
https://arstechnica.com/gadgets/2024/04/amazon-ends-ai-powered-store-checkout-which-needed-1000-video-reviewers/
AI models learn from data. If this data reflects existing societal biases or underrepresents certain racial groups, the model will inherit those biases.
@user18658159 AKA Microsoft’s Tay.
https://en.m.wikipedia.org/wiki/Tay_(chatbot)
I have no mouth and I must scream.
Written by Harlan Ellison, published in 1967.
There was a supercomputer AI called Artificial Mind that was the result of all AI on the planet merging. And it hated us. And it reduced the Human race down to five people that it kept around to torment.
So there’s that.
@PocketBrain yeah people don’t realize A.I. is not a new thing, just the super-powerful chips and huge database of all of us called the internetz, which didn’t in those days.
Might also look up “the Adolescence of P1” I think it was.
@PocketBrain One of my favorite book scenes is in Asimov’s Foundation. The Imperial Library on Trantor had “reader stations” where book/document texts could be read on a screen. If a reader were to rest his eyes on a single word or phrase for a couple of seconds, it would “jump” to a definition of the word or a longer explanation of the topic. So: hypertext, as described by Asimov, circa 1950.
@pmarin @PocketBrain if you think about AI as “giving power of control to the lowest common denominator”, isn’t that just politicians?
@pakopako @PocketBrain eek, you are right, but that is probably inevitable. Our education system has probably slipped so much that almost none, especially the politicians, who cut class to go smoke behind the football field, have no idea what an LCD is, and I don’t mean your screen.
@pmarin @PocketBrain
https://sproutsschools.com/cipollas-5-laws-of-human-stupidity/
TL;DR - humans can be broken down into 4 types (beneficial, sacrificial, hoarding, and stupid)
You want to spur on those who can figure out a way to create something for both you and me, kind of okay with those that believe in self-sacrifice so at least someone gets theirs, be mindful of those who will betray you to suit their own needs, and avoid the unpredictably of a stupid person at all costs.
@pakopako @PocketBrain Long before semiconductor chips, this was shown by the Greeks and then Romans. amazing geniuses and inventors. Political corruption. Occasional assassination “et Tu, Brute” allegedly. So you are saying the A.I. will both advance and degrade our civilization at a much faster rate? Oh the upcoming “dark ages” will be fun.
@pmarin @PocketBrain you know the old robot joke
“My processor is newer, so that means I get to make mistakes faster!”
Skynet
@ebatch Elon is up to about 10 launches a week now, each with 20-25 starbots. I think when Jeff Bezos launched his first batch for his own network, Elon ramped up his launches. Always wonder if the starbots will have space-to-space combat and the truth is we would have no idea what is happening up there.
There’s no AI yet, Large Language Models are just glorified autocomplete, and useless without large scale theft of intellectual property for training.
Currently I think the biggest problem is they lie with same certainty that they have when getting it right. What use are they really, when you have to double check their output? And since this is a well known problem, why don’t the LLM makers have a verification step built in?
@blaineg I agree with you; not yet up to self-evolving self-conscious A.I. (is that even possible? Has been debated for at least 50 years)
Also maybe we could have A.I. check the checkers? Brings up the classic security paradox, rephrased: “who will check the checkers?”
@blaineg @pmarin I find it “interesting” that the “are you a human” test grids are AI-generated and checked. So we have to convince a computer that we are human - the opposite of passing a Turing test. I don’t think Alan Turing ever saw that coming.
On our topic, look up the movie Dark Star. I didn’t realize until recently it was directed by John Carpenter. It is kind of a hippie-druggie-sci-fi-comedy. (It was the 80’s)
Aside from entertaining combat with a giant alien beach ball, the main point was the “smart bomb,” which was told its purpose was to explode. But it was a glitch, so they tried to stop it (don’t want to spoil more than that). They (commander and the smart bomb) eventually discuss “phenomenology,” and 40 years later I’m still not sure what that is.
@pmarin Okay, I think I’m going to have to watch that.
@Kyeh It would be very “meta” but now I’m thinking I should ask the A.I. to teach me phenomenology.
@pmarin Or, you know, just read a book.
@Kyeh @pmarin
/youtube the tick handy read a book
@pmarin @Kyeh I remember watching Dark Star (in the 70’s) at the midnight movie at Fifth Avenue Cinema in Portland (high, of course). It’s available online (free, multiple sources) - my wife and I watched it recently - still fun, but not as much as I recalled from my original viewing. Pretty good/fun SFX, given the shoestring budget.