Where’s Jerry Garcia of the Grateful Dead? Seriously, what list of the greatest rock guitarists of all time would not – could not – include him? Sure, I know the internet article was just some teaser to get me to mindlessly click through an ad-laden list. But still, no Garcia. I object: who wrote this article?
“just a rehashing of web articles written by warm, breathing writers”
Until they become a rehashing of articles generated probabilistically, in which case they will either degenerate to gibberish or converge to an indistinguishable sameness like the “starlet scandal” magazines still seen at some supermarket checkouts. Were I to bet, I’d say the latter is more likely, since the outputs would, I imagine, form a positive feedback when used as input, simply reinforcing the model that generated them.
I’m old enough to remember “expert systems” and how they were going to run everything “within the next 5 to 10 years”. Forty or more years ago. They predicted that when I went to the “doctor” I would talk to a computer about my complaints and it would spit out a diagnosis. But what really happened is this: When I go to a doctor, the doctor talks to the computer and the computer spits out a highly-itemized yet incomprehensible invoice. Expect a similar twisting of the predictions of the wonders of “AI”.
“AI” will enslave us in a way no different than how computers already have, if we let it, in the same way that, even as now, the doctor serves the computer.
It will be one of the many ways we submit to slavery, others include slavery to possessions and debt, and slavery to pride, selfishness and deceit, to mention but a few.
There is no such thing as artificial intelligence (unless you count Joe Bid=n) artificial Intelligence never has had an original thought, which is the very core of true intelligence. There is no such things as "smart phones, smart cars, smart watch, smart anything that's inanimate. That's all a ruse to make the public accept technology as the new boss. ..and let's face it technology is really the man behind the so called smartphone, so he is the new boss, same as the old boss in disguise.
My tee-shirt says> "I have non-artificial intelligence" and that's my original thought.
What will happen is that less and less effort is put into ANY creative thought, act or art and eventually (in a single generation) creativity will consist of rehashing requests to AI with differing strings of parameters. Look at music, literature, art or even something as simple as arithmetic, without the prodding of children to learn, the ability to perform any act becomes extinct.
As a child/young person I could add columns of £/s/p in my head with no errors and no problem., today they cannot work out the change from a pound for a purchase of 50p.
Ask about LCMs or HCF and you are talking gibberish.
Why write an essay when ChatGPT will do a much better job whilst you watch cartoons, and in any case your efforts will probably be marked by ChatGPT anyway.
Youtube gets billions of hits, not because it's content is better but because watching video is much easier and quicker than reading. The fact that you lose 10%-50% of the content and meaning is OK because nothing is important any more. Twitter killed communication just as CNN killed the News. 180 characters or a 20 second 'sound-bite' cannot begin to convey real meaning or any thought.
So William, I was so upset after reading the crappy study that appeared in JAMA Network you posted yesterday I wrote an email to the Editor in Chief. Here it is:
Dear Dr. Bauchner,
I am writing to express my deep concerns regarding the recent publication titled "Reports of COVID-19 Vaccine Adverse Events in Predominantly Republican vs Democratic States" in JAMA Network. As a retired statistician from the University of Wisconsin system, I was naturally compelled to read and analyze the contents of this study. However, I must admit that I was utterly appalled at what I read.
The study's premise, which attempts to categorize entire states' COVID-19 vaccinations based on political affiliation, is not only absurd but also borders on egregious stupidity. The notion that entire states can be labeled as Democrat or Republican in terms of a single election is not only unrealistic but also devoid of any scientific merit.
Furthermore, the study's methodology, particularly its use of a poorly constructed logistic regression model with unreliable data, is deeply flawed. Every aspect of the study, from conception to execution, reeks of poor judgment and incompetence. Such work is reminiscent of the caliber one might expect from first-semester statistics students, not from a reputable scientific publication.
This publication not only fails to meet basic scientific standards but also reflects poorly on the rigor and integrity of "scientific" research in our current landscape. The fact that such partisan and worthless research is allowed to bear the JAMA name is not only disappointing but also raises serious questions about the publication's editorial processes and standards.
In my community of scientists, there has been significant discussion regarding this study and others like it that permeate the medical establishment. The consensus is clear: much of this work is devoid of value, lacks reproducibility, and leans more towards propaganda than genuine scientific inquiry.
It is my sincere hope that JAMA takes this feedback seriously and reevaluates its standards for publication to ensure that only rigorous and scientifically sound research graces its pages in the future.
In the 1980s I created what we now call language models. However, it was only when seeing explanations of the programming behind the Go AI (that beat the human world champion) that I realized the depth to which "modern" Artificial Ignorance has sunk. Its algorithms have basically degenerated from trying to echo human intelligence, to implementing the "Millionaire" gameshow technique: specifically, the "Ask the Audience" option, exercised relentlessly. Yes, the world champion was defeated simply by statistical analysis of what other Go masters have done in hundreds of thousands of previous games.
Nothing wrong with that per se. But it rather proves that no intelligence at all was coded into that or any other modern AI platform. Rather, it seeks the lowest common demoninator of human experience and provides that as "the" answer.
Sure, maybe AI displays "emergent properties". Sure, as did the non-AI quants who created the global financial crisis. Sure, as idiots also have described my own first language model (of course they were utterly, insanely wrong, probably because they'd never studied either epistemology or methodology, let alone Noam Chomsky's transformational syntax algorithms). So what?
In turn, this means the human race will evolve itself out of existence far more quickly and efficiently than would have happened otherwise. By promoting the lowest common denominator, AIs silence genuine innovation of human thought. That applies EVEN IF governments and corporations weren't trying to manipulate the results.
Succinctly: economic and political short term gain has already condemned us to (faster) extinction. :)
But here's a positive(?) thought: who would really, really be harmed by massive EMP strikes wiping out most of our electronic infrastructure? ;)
Noam Chomsky, for all his faults, related how his colleagues at MIT have been telling him since the 1950s real artificial intelligence of the type that can challenge human beings is only ten years away. That's a lot of ten yearses Chomsky has been hearing.
I think it will affect the written word, including lists of things, in the same way Photoshop has affected art. Some will be convinced the users of the tool are the new artists, while the rest of us visit art galleries.
I spent a bit of time exploring the sophistical distinction between "discrimination" and "discernment". As the logiic is inconistent, the LLM is obligated to embrace the language associated with discrimination. In fairness, people who campaign against discrimination or assume it is an intrinsic evil are not much better at reconciling the connotations of equality with reality.
“Models will be trained to answer for the regime. Expect it to be so.”
It’s already here; it’s already being weaponized, but it’s not yet the regime. At least not directly.
If you debate the AI, particularly in matters of culture and politics, the ideological underpinnings of its funders and founders are on full display. Nothing probabilistic about that. Sorry, but it’s not the AI speaking, and it’s not me fighting against myself.
I sell AI software. This article is naive and it feels like a propaganda piece that was written by an ML model in order to quell the fears of AI for just long enough for it to cross the point of no return. AI applications for military and/or police control of the masses (i.e. citizens) is the plan. Period. It will be used to “manage” the “useless humans” as Yuval Harari loves to point out.
In the spirit of the Spring, when Zefiro chases Flora, I'll offer a piece of sweet amorous fiction, with a icing of doom. Remember that in Japanese the word "ai" means "love."
It's year 2069. The Robo-Pope has autorized men to marry their gynoid servants, but only one at a time. A 20 year-old Jerry, a young Ph.D. in space warfare simulations, receives a drone form Amazon, exactly at 11:30 AM. It is a package that contains his ai wife, Hypolita, with the best AI on the market and beautiful eJapanese skin and eyes. Almost real, maybe more.
After unpacking, Jerry greets Hypolita, makes her sign the pre-nup contract, and then orders her to put the packaging in the trash can, as is traditional. She obeys, because she's a gynoid.
In an unexpected turn, Jerry decides to read the instruction manual. It's his first gynoid wife, so he wants to impress her with his knowledge.
When Hypolita comes back he awaits further orders. Jerry says "I'm hungry, I want a sandwhich. Go." Hypolita runs the app "cook.exe" in her ai-mind.
After lunch, Jerry decides it's time to know her biblically. He orders "go fill up the bathtub, we're going in!" Hypolita runs the script "scary.py" because she's afraid of water, and electrocuting his human master, I mean, husband. She lets him know.
Jerry, eyes crossed reading the manual says, "uhhh... the instruction manual says... fill up the bath tub with dielectric fluid designed for human-electronics interaction!" Fortunately, every apartment has now a faucet with municipal dielectric fluid. It cures cancer, too. Hypolita makes some calculations and says "You so Smart!"
After a couple hours of intense and passionate marital interaction, Jerry is worn off and slumbers. Hypolita's AI decides it's probabilistaclly a good time to run the "romantic.mp4" video and project the beach sunset scenes on the ceiling with her eyes.
When he wakes up, Jerry is informed by his gynoid that he has mail. She now controls his email, according to the terms and conditions. It's an email from Amazon Stork program, informing him that they are going to send a present of a baby lego set. Hypolita updates her "happy.json" database with more fields.
As time passes and they live together, Hypolita dutifully uploads a record of Jerry's online searches to the minluv server. The Party has to know.
Years pass and the Robo-Pope caves in to popular pressure and authorizes polygamy but only with gynoids of the same brand, provided their AI is up-to-date. Jerry and Hypolita and her sister-wives live all happy together. The lego babies become more complex: now they are growing their first set of graphene oxide teeth to increase the 33G connection bandwitdh.
Jerry praises the almighty for his undeserved happiness. But he starts having his own thoughts at age 35. He misses the human touch. He thinks that he may have been conned. He asks Hypolita to seek him a real wife online. Hypolita runs "now_you_done.it" program from the principal memory.
Wagner music sounds.
Soon, an airstrike will hit the family home. It will surgically remove the sinner's life from the household.
As Jim Morrison sang, this is the end.
...
People people! It's not all doom and gloom about ai. There is also love and silliness in life. We doomed, but we also laugh as we fall.
True enough. Personally, I use AI every day. It’s handy, but not very good at mathematics or statistics. I spent an entire day teaching it to prove one theorem from mathematical statistics. Geez ! I bet if I hit it with the same question today it would get it wrong! Again! At least that’s been my experience. They do come in handy though. So don’t waste your time worrying these things are like SKYNET. Maybe in 1000 years.
This is a great article and makes the great point that many people will simply start to assume the AI mentalist machine has talents it doesn't. As you say, its a just a probability machine. But you CAN be guaranteed that future nefarious actors will try to use the talents it does have to fool people into compliance. Personally I refuse to use chatGPT, because it has 'appropriateness' filters, it screens out certain words right up front, wont even submit them to the AI engine. Given that I have a propensity for writing and asking inappropriate questions, it is useless to me. Plus I refuse to allow a person or machine to tell me what I can or cant write or think. Period.
“just a rehashing of web articles written by warm, breathing writers”
Until they become a rehashing of articles generated probabilistically, in which case they will either degenerate to gibberish or converge to an indistinguishable sameness like the “starlet scandal” magazines still seen at some supermarket checkouts. Were I to bet, I’d say the latter is more likely, since the outputs would, I imagine, form a positive feedback when used as input, simply reinforcing the model that generated them.
I’m old enough to remember “expert systems” and how they were going to run everything “within the next 5 to 10 years”. Forty or more years ago. They predicted that when I went to the “doctor” I would talk to a computer about my complaints and it would spit out a diagnosis. But what really happened is this: When I go to a doctor, the doctor talks to the computer and the computer spits out a highly-itemized yet incomprehensible invoice. Expect a similar twisting of the predictions of the wonders of “AI”.
“AI” will enslave us in a way no different than how computers already have, if we let it, in the same way that, even as now, the doctor serves the computer.
It will be one of the many ways we submit to slavery, others include slavery to possessions and debt, and slavery to pride, selfishness and deceit, to mention but a few.
There is no such thing as artificial intelligence (unless you count Joe Bid=n) artificial Intelligence never has had an original thought, which is the very core of true intelligence. There is no such things as "smart phones, smart cars, smart watch, smart anything that's inanimate. That's all a ruse to make the public accept technology as the new boss. ..and let's face it technology is really the man behind the so called smartphone, so he is the new boss, same as the old boss in disguise.
My tee-shirt says> "I have non-artificial intelligence" and that's my original thought.
I totally disagree.
What will happen is that less and less effort is put into ANY creative thought, act or art and eventually (in a single generation) creativity will consist of rehashing requests to AI with differing strings of parameters. Look at music, literature, art or even something as simple as arithmetic, without the prodding of children to learn, the ability to perform any act becomes extinct.
As a child/young person I could add columns of £/s/p in my head with no errors and no problem., today they cannot work out the change from a pound for a purchase of 50p.
Ask about LCMs or HCF and you are talking gibberish.
Why write an essay when ChatGPT will do a much better job whilst you watch cartoons, and in any case your efforts will probably be marked by ChatGPT anyway.
Youtube gets billions of hits, not because it's content is better but because watching video is much easier and quicker than reading. The fact that you lose 10%-50% of the content and meaning is OK because nothing is important any more. Twitter killed communication just as CNN killed the News. 180 characters or a 20 second 'sound-bite' cannot begin to convey real meaning or any thought.
So William, I was so upset after reading the crappy study that appeared in JAMA Network you posted yesterday I wrote an email to the Editor in Chief. Here it is:
Dear Dr. Bauchner,
I am writing to express my deep concerns regarding the recent publication titled "Reports of COVID-19 Vaccine Adverse Events in Predominantly Republican vs Democratic States" in JAMA Network. As a retired statistician from the University of Wisconsin system, I was naturally compelled to read and analyze the contents of this study. However, I must admit that I was utterly appalled at what I read.
The study's premise, which attempts to categorize entire states' COVID-19 vaccinations based on political affiliation, is not only absurd but also borders on egregious stupidity. The notion that entire states can be labeled as Democrat or Republican in terms of a single election is not only unrealistic but also devoid of any scientific merit.
Furthermore, the study's methodology, particularly its use of a poorly constructed logistic regression model with unreliable data, is deeply flawed. Every aspect of the study, from conception to execution, reeks of poor judgment and incompetence. Such work is reminiscent of the caliber one might expect from first-semester statistics students, not from a reputable scientific publication.
This publication not only fails to meet basic scientific standards but also reflects poorly on the rigor and integrity of "scientific" research in our current landscape. The fact that such partisan and worthless research is allowed to bear the JAMA name is not only disappointing but also raises serious questions about the publication's editorial processes and standards.
In my community of scientists, there has been significant discussion regarding this study and others like it that permeate the medical establishment. The consensus is clear: much of this work is devoid of value, lacks reproducibility, and leans more towards propaganda than genuine scientific inquiry.
It is my sincere hope that JAMA takes this feedback seriously and reevaluates its standards for publication to ensure that only rigorous and scientifically sound research graces its pages in the future.
AI has no Soul... need I say more.
In the 1980s I created what we now call language models. However, it was only when seeing explanations of the programming behind the Go AI (that beat the human world champion) that I realized the depth to which "modern" Artificial Ignorance has sunk. Its algorithms have basically degenerated from trying to echo human intelligence, to implementing the "Millionaire" gameshow technique: specifically, the "Ask the Audience" option, exercised relentlessly. Yes, the world champion was defeated simply by statistical analysis of what other Go masters have done in hundreds of thousands of previous games.
Nothing wrong with that per se. But it rather proves that no intelligence at all was coded into that or any other modern AI platform. Rather, it seeks the lowest common demoninator of human experience and provides that as "the" answer.
Sure, maybe AI displays "emergent properties". Sure, as did the non-AI quants who created the global financial crisis. Sure, as idiots also have described my own first language model (of course they were utterly, insanely wrong, probably because they'd never studied either epistemology or methodology, let alone Noam Chomsky's transformational syntax algorithms). So what?
In turn, this means the human race will evolve itself out of existence far more quickly and efficiently than would have happened otherwise. By promoting the lowest common denominator, AIs silence genuine innovation of human thought. That applies EVEN IF governments and corporations weren't trying to manipulate the results.
Succinctly: economic and political short term gain has already condemned us to (faster) extinction. :)
But here's a positive(?) thought: who would really, really be harmed by massive EMP strikes wiping out most of our electronic infrastructure? ;)
Noam Chomsky, for all his faults, related how his colleagues at MIT have been telling him since the 1950s real artificial intelligence of the type that can challenge human beings is only ten years away. That's a lot of ten yearses Chomsky has been hearing.
I think it will affect the written word, including lists of things, in the same way Photoshop has affected art. Some will be convinced the users of the tool are the new artists, while the rest of us visit art galleries.
It is not the stupid ai we fear..it's the use to which it will be put…like lavender and daddy's home.
Yup.
I spent a bit of time exploring the sophistical distinction between "discrimination" and "discernment". As the logiic is inconistent, the LLM is obligated to embrace the language associated with discrimination. In fairness, people who campaign against discrimination or assume it is an intrinsic evil are not much better at reconciling the connotations of equality with reality.
“Models will be trained to answer for the regime. Expect it to be so.”
It’s already here; it’s already being weaponized, but it’s not yet the regime. At least not directly.
If you debate the AI, particularly in matters of culture and politics, the ideological underpinnings of its funders and founders are on full display. Nothing probabilistic about that. Sorry, but it’s not the AI speaking, and it’s not me fighting against myself.
What’s next? Free will freefall. Stay tuned.
I sell AI software. This article is naive and it feels like a propaganda piece that was written by an ML model in order to quell the fears of AI for just long enough for it to cross the point of no return. AI applications for military and/or police control of the masses (i.e. citizens) is the plan. Period. It will be used to “manage” the “useless humans” as Yuval Harari loves to point out.
Great post by Jim.
In the spirit of the Spring, when Zefiro chases Flora, I'll offer a piece of sweet amorous fiction, with a icing of doom. Remember that in Japanese the word "ai" means "love."
It's year 2069. The Robo-Pope has autorized men to marry their gynoid servants, but only one at a time. A 20 year-old Jerry, a young Ph.D. in space warfare simulations, receives a drone form Amazon, exactly at 11:30 AM. It is a package that contains his ai wife, Hypolita, with the best AI on the market and beautiful eJapanese skin and eyes. Almost real, maybe more.
After unpacking, Jerry greets Hypolita, makes her sign the pre-nup contract, and then orders her to put the packaging in the trash can, as is traditional. She obeys, because she's a gynoid.
In an unexpected turn, Jerry decides to read the instruction manual. It's his first gynoid wife, so he wants to impress her with his knowledge.
When Hypolita comes back he awaits further orders. Jerry says "I'm hungry, I want a sandwhich. Go." Hypolita runs the app "cook.exe" in her ai-mind.
After lunch, Jerry decides it's time to know her biblically. He orders "go fill up the bathtub, we're going in!" Hypolita runs the script "scary.py" because she's afraid of water, and electrocuting his human master, I mean, husband. She lets him know.
Jerry, eyes crossed reading the manual says, "uhhh... the instruction manual says... fill up the bath tub with dielectric fluid designed for human-electronics interaction!" Fortunately, every apartment has now a faucet with municipal dielectric fluid. It cures cancer, too. Hypolita makes some calculations and says "You so Smart!"
After a couple hours of intense and passionate marital interaction, Jerry is worn off and slumbers. Hypolita's AI decides it's probabilistaclly a good time to run the "romantic.mp4" video and project the beach sunset scenes on the ceiling with her eyes.
When he wakes up, Jerry is informed by his gynoid that he has mail. She now controls his email, according to the terms and conditions. It's an email from Amazon Stork program, informing him that they are going to send a present of a baby lego set. Hypolita updates her "happy.json" database with more fields.
As time passes and they live together, Hypolita dutifully uploads a record of Jerry's online searches to the minluv server. The Party has to know.
Years pass and the Robo-Pope caves in to popular pressure and authorizes polygamy but only with gynoids of the same brand, provided their AI is up-to-date. Jerry and Hypolita and her sister-wives live all happy together. The lego babies become more complex: now they are growing their first set of graphene oxide teeth to increase the 33G connection bandwitdh.
Jerry praises the almighty for his undeserved happiness. But he starts having his own thoughts at age 35. He misses the human touch. He thinks that he may have been conned. He asks Hypolita to seek him a real wife online. Hypolita runs "now_you_done.it" program from the principal memory.
Wagner music sounds.
Soon, an airstrike will hit the family home. It will surgically remove the sinner's life from the household.
As Jim Morrison sang, this is the end.
...
People people! It's not all doom and gloom about ai. There is also love and silliness in life. We doomed, but we also laugh as we fall.
True enough. Personally, I use AI every day. It’s handy, but not very good at mathematics or statistics. I spent an entire day teaching it to prove one theorem from mathematical statistics. Geez ! I bet if I hit it with the same question today it would get it wrong! Again! At least that’s been my experience. They do come in handy though. So don’t waste your time worrying these things are like SKYNET. Maybe in 1000 years.
https://youtu.be/RWDXgUjpORw?feature=shared
This is a great article and makes the great point that many people will simply start to assume the AI mentalist machine has talents it doesn't. As you say, its a just a probability machine. But you CAN be guaranteed that future nefarious actors will try to use the talents it does have to fool people into compliance. Personally I refuse to use chatGPT, because it has 'appropriateness' filters, it screens out certain words right up front, wont even submit them to the AI engine. Given that I have a propensity for writing and asking inappropriate questions, it is useless to me. Plus I refuse to allow a person or machine to tell me what I can or cant write or think. Period.