"If you tell it to “Write a response to an email from my homeowners association, explaining that I am not going to stop performing simulated oral sex on the statue of Lucy Lawless that I bought from Wish.com and put out in front of my bedroom window,”
And THIS is why I read your Substack, Jack.
Feel free to sign me up for another year's subscription.
I really should tell you some time about when a good friend of mine, one night at the restaurant he managed, got an irate call from the father of one of the comely teenaged lasses that waited his tables, that began with the words, "YOU WANNA TELL ME WHY YOUR GUYS WERE SEXUALLY HARASSING MY DAUGHTER WITH AN INFLATABLE FUCK SHEEP???!!!"
When a friend of mine shared the ChatGTP-generated e-mail telling someone politely that he stinks, and another telling a graphic designer very nicely that the logo he created is shit, I replied – and realised – that the time will come, and it will not be long, when using some LLM to answer your work (let alone personal) communication will be viewed as insult.
The time might actually come when people, tired to death by reading long litanies of LLMs, will just whip up their old phones, call the other party and just plain tell them that their idea is totally fucking stupid, that they won't spend their time on another moronic conference call because it's pure bullshit and that the last time the other fucker was in the office, he actually stank.
And said fucker will be thankful, although he will also tell the first one that he's a cocksucker, his brief for the logo looked like a drunk gorilla wrote it and that his hairstyle could only be popular in 1990s, among people who were on drugs.
Then, they maybe let LLMs continue exchange pleasant-worded nasty e-mails and will go sort it out over a beer, or ten.
I'm probably too much of an optimist but, well, a man can dream!
On the other hand, if some of the recent college grads who apply to positions that I have posted would use it to put together their resumes so I wouldn’t have to wade through something that is an embarrassment to higher education, it would feel like a bit of a relief.
The last time I hired somebody was probably ’16. Errors and poorly written resumes were always an easy first cut. If they’re using this, they might make the interview stage— waste of your time.
I contemplated doing freelance resume work; but I couldn’t bring myself to do it. I didn’t want to be some meathead’s smoke screen.
That is exactly what I used to do. The sad thing is that we have such a shortage in our field that I now often have to overlook that in many cases in order to even find potential candidates to interview.
So onboarding now includes enough English training--writing, not reading--for the candidate to be able to construct an Email or process documentation at a high school-graduate level! Or should!
I’m going to have to reread this one tomorrow as I am too filled with joy to be able to ponder deeper meanings. I have discovered that I can respond to every single piece of correspondence from the petty tyrants of my HOA in iambic pentameter. Oh joyous day!
I only laugh at things that're antisocial, dangerous, individualistic and/or well said, and the idea of some perv doing a statue of some celebrity babe in full view of the neighborhood...
I haven't laughed that hard since I found out they're remaking "The Toxic Avenger" with Peter Dinklage in the title role.
Jack how do I get the LLM to write my messages to the HOA, they get offended to all my messages written in the voice of David Berry maybe a computer generated message would be less upsetting to their idea that my communicatioins creates a hostile work (???) place.
What do you have against David Gelerntner? I like how Gelerntner told Time magazine to fuck off when they wanted him to do a point-counterpoint with Ted K. 'You want me to debate the guy who crippled and tried to kill me?!!"
Nothing at all. I'm just saying that history will conclude that Ted was right. His methods were ridiculous, but what would you expect from an academic?
From what I can tell this is the entire thrust of the "value add" apps that every fucking company from here to Timbuktu has to have.
Though I can't say that's an original thought on my part I find it true. As ever the least capable are most damaged by these "helpful" technological intermediaries.
My God, some of these things are over 300 MB! And they’re constantly being updated! I don’t have an unlimited data plan on my cell phone, and a 3mpbs down DSL (or whatever it is), so it takes a bit of time to download those suckers!
I suppose it’s just more spyware, but at least with Apple, I think they’re the least objectionable in terms of selling your soul! I’m trying to think if I have a Google account--I don’t think so!
If you aren’t paying for a product, YOU are the product!
I look forward to the responses of your autistic readers(most of us)
Everything is too long didn’t read bullshit. My resume, if i ever needed one again, should say. I complete 200-300 personal tax returns and 100-200 business tax returns. I also do tax planning. Id have turn that into 2 pages of bullshit to find a job. It should be obvious if i prepare that many tax returns, I’m at least somewhat proficient in the software to prepare those tax returns. Maybe i have a bullet point for the additional tax returns i review and i know how to use a spreadsheet. We are currently in the process of trying to hire a couple tax people. I cannot ask a straightforward question and get a straightforward answer because most of the people who make it to partner in my profession are bullshitters and salesman and most of the interviewees are trained for this so all the resumes and responses are fluff. I am neither a bullshitter or a salesman. It all reads like some whiney bitch on reddit because it was trained form whiney bitches on reddit
I too have a short attention span but i do find the use of LLMs to condense things to summarize the key parts, like ToS and privacy policies, or a long scientific article
There's a lot of hand-wringing that 'AI' is going to 'take all our jobs'. What it really means, is that if it really can do your job then your job is probably not needed at all in the first place. 'Learn to code'.
I asked ChatGPT two basic questions about a commodity market and an adjacency the other day, related to my job, and it gave me an answer to both like a kid who read maybe the headings in last night’s assigned readings and nothing more. The information was outdated, irrelevant, incorrect, and so generic as to be useless. My job is safe for a while longer.
“The price for [food commodity X] has declined dramatically over the last 9 months, why has demand for this commodity not increased?”
“Why did [large group of businesses] all reduce their purchase of [commodity] suddenly in April?”
In my mind the concern is less that it really can do my job, and more that it can in the short term make some senior manager think that it can do my job because he has no idea what I really do. The end result for me is the same.
Just an aside, but stuff like this is why you DON'T want X-ray vision.
You'd be rolling down the street, on your way to the store, casually looking into people's houses and you'd see some naked fat dude in his bedroom fucking a five-cheese burrito.
It just seems to be another way for people to get more lazy and stupid. And frankly, there’s already enough of both around.
I’m still astounded that our esteemed VP had some sort of AI summit. The mind boggles. How about asking for a summary of one of her “speeches”? Can a computer commit suicide?
Thirty years ago my favorite prof showed us the latest version of a 2D electromagnetic modeling package and commented that it did the job that eight guys had done all day, prior to electronic computing. It's a pity that Spiro Agnew wasn't assigned the task of crafting a response to the automation of Finite Element Analysis. The world needed the laughs!
"I predict that in the future, computers will be twice as powerful, 10,000 times as large and only the five richest kings of Europe will be able to afford them."
I accomplished something today, though. I know from college speech class that “dyadic” communication is between two people, but I didn’t know what the equivalent is if you’re talking to an inanimate object, like a computer, when you’re instructed to “converse” with a computer on the other end of the phone! It’s “monadic!” 👍
“An order that can be misunderstood, will be misunderstood.” - Napoleon
"Always remember that it is impossible to speak in such a way that you cannot be misunderstood: there will always be some who misunderstand you.” - Karl Popper
Does Popper's comment apply to LLM? Is there a way, short of using formal programming languages, to make sure that a computer fully understands the input?
I thought your nom de plume was used in Jack’s “Car And Streeter” stuff at the insurance company which shall not be named!
I thought that writer was driving something nuclear powered, lost control, and hit an elementary school, then a fission reaction spontaneously occurred, and took out the surrounding subdivision! 😬
"If you tell it to “Write a response to an email from my homeowners association, explaining that I am not going to stop performing simulated oral sex on the statue of Lucy Lawless that I bought from Wish.com and put out in front of my bedroom window,”
And THIS is why I read your Substack, Jack.
Feel free to sign me up for another year's subscription.
There may be a market for a life-size, anatomically correct Lucy with which to perform described acts, but Jack would hardly be the target demo.
I really should tell you some time about when a good friend of mine, one night at the restaurant he managed, got an irate call from the father of one of the comely teenaged lasses that waited his tables, that began with the words, "YOU WANNA TELL ME WHY YOUR GUYS WERE SEXUALLY HARASSING MY DAUGHTER WITH AN INFLATABLE FUCK SHEEP???!!!"
How would ChatGPT answer that?
Probably an indignant diatribe about assuming the animal's gender.
https://www.youtube.com/watch?v=tFOMtRF6MIk Batwing.
"Feel free to sign me up for another year's subscription. "
To adapt Buckley: "_You_ sign yourself up!"
Not gonna lie I scrolled through this post "ok more computer stuff." Then I saw in the comments Lucy Lawless being discussed.... *now* we're talking
When a friend of mine shared the ChatGTP-generated e-mail telling someone politely that he stinks, and another telling a graphic designer very nicely that the logo he created is shit, I replied – and realised – that the time will come, and it will not be long, when using some LLM to answer your work (let alone personal) communication will be viewed as insult.
The time might actually come when people, tired to death by reading long litanies of LLMs, will just whip up their old phones, call the other party and just plain tell them that their idea is totally fucking stupid, that they won't spend their time on another moronic conference call because it's pure bullshit and that the last time the other fucker was in the office, he actually stank.
And said fucker will be thankful, although he will also tell the first one that he's a cocksucker, his brief for the logo looked like a drunk gorilla wrote it and that his hairstyle could only be popular in 1990s, among people who were on drugs.
Then, they maybe let LLMs continue exchange pleasant-worded nasty e-mails and will go sort it out over a beer, or ten.
I'm probably too much of an optimist but, well, a man can dream!
Its like those cursive personilzed notes i get from realtors that are obviously mail merged bullshit
Yes! Exactly!
I guarantee I would rip someone a new one if I was sent a response written by that crap at work. If I recognized it. And I wasn’t retired.
Come to think of it, since many can’t string a coherent sentence together— if suddenly they could, I would know it was this BS doing the talking!
On the other hand, if some of the recent college grads who apply to positions that I have posted would use it to put together their resumes so I wouldn’t have to wade through something that is an embarrassment to higher education, it would feel like a bit of a relief.
The last time I hired somebody was probably ’16. Errors and poorly written resumes were always an easy first cut. If they’re using this, they might make the interview stage— waste of your time.
I contemplated doing freelance resume work; but I couldn’t bring myself to do it. I didn’t want to be some meathead’s smoke screen.
That is exactly what I used to do. The sad thing is that we have such a shortage in our field that I now often have to overlook that in many cases in order to even find potential candidates to interview.
Wow. That is incredibly sad.
So onboarding now includes enough English training--writing, not reading--for the candidate to be able to construct an Email or process documentation at a high school-graduate level! Or should!
We are truly fucked, aren’t we?! 🤦♂️
Bingo!
I wrote resumes for 10 years. You get into a piecemeal mindset that never lets you relax.
Hope you have brushed up on your Voight-Kampff test questions.
a Blade Runner reference in the column AND the comments. bravo
That would have been the obvious outcome back in 1925, but today's humans have been meekly adjusted to the needs of the systems.
I’m going to have to reread this one tomorrow as I am too filled with joy to be able to ponder deeper meanings. I have discovered that I can respond to every single piece of correspondence from the petty tyrants of my HOA in iambic pentameter. Oh joyous day!
I only laugh at things that're antisocial, dangerous, individualistic and/or well said, and the idea of some perv doing a statue of some celebrity babe in full view of the neighborhood...
I haven't laughed that hard since I found out they're remaking "The Toxic Avenger" with Peter Dinklage in the title role.
Call me an elf one more time... oooh, he’s an angry elf...
Jack how do I get the LLM to write my messages to the HOA, they get offended to all my messages written in the voice of David Berry maybe a computer generated message would be less upsetting to their idea that my communicatioins creates a hostile work (???) place.
If i had to write to an hoa, my messages would be more likely to be from teddy k, not the kennedy, than chatgpt. Sounds awful.
Who would have thought the sanest man in America was living in the woods
What do you have against David Gelerntner? I like how Gelerntner told Time magazine to fuck off when they wanted him to do a point-counterpoint with Ted K. 'You want me to debate the guy who crippled and tried to kill me?!!"
Nothing at all. I'm just saying that history will conclude that Ted was right. His methods were ridiculous, but what would you expect from an academic?
"It can only add unnecessary complexity"
From what I can tell this is the entire thrust of the "value add" apps that every fucking company from here to Timbuktu has to have.
Though I can't say that's an original thought on my part I find it true. As ever the least capable are most damaged by these "helpful" technological intermediaries.
We can try "valued subtract."
It's pretty much a blanket description of my whole work history.
I think it’s apps in general!
My God, some of these things are over 300 MB! And they’re constantly being updated! I don’t have an unlimited data plan on my cell phone, and a 3mpbs down DSL (or whatever it is), so it takes a bit of time to download those suckers!
I suppose it’s just more spyware, but at least with Apple, I think they’re the least objectionable in terms of selling your soul! I’m trying to think if I have a Google account--I don’t think so!
If you aren’t paying for a product, YOU are the product!
We invented the internet so we wouldn’t have to have apps and theres these Things called bookmarks to quickly access sites. I loathe most apps.
I look forward to the responses of your autistic readers(most of us)
Everything is too long didn’t read bullshit. My resume, if i ever needed one again, should say. I complete 200-300 personal tax returns and 100-200 business tax returns. I also do tax planning. Id have turn that into 2 pages of bullshit to find a job. It should be obvious if i prepare that many tax returns, I’m at least somewhat proficient in the software to prepare those tax returns. Maybe i have a bullet point for the additional tax returns i review and i know how to use a spreadsheet. We are currently in the process of trying to hire a couple tax people. I cannot ask a straightforward question and get a straightforward answer because most of the people who make it to partner in my profession are bullshitters and salesman and most of the interviewees are trained for this so all the resumes and responses are fluff. I am neither a bullshitter or a salesman. It all reads like some whiney bitch on reddit because it was trained form whiney bitches on reddit
I too have a short attention span but i do find the use of LLMs to condense things to summarize the key parts, like ToS and privacy policies, or a long scientific article
There's a lot of hand-wringing that 'AI' is going to 'take all our jobs'. What it really means, is that if it really can do your job then your job is probably not needed at all in the first place. 'Learn to code'.
If your job wasn't done by a cat or a boa constrictor in a Richard Scarry book, then you're probably not that important.
Comment of the YEAR
Lowly Worm on a business card would probably be less sus than “peon.”
Watch your step, Mr Rabbit
Those books were pretty cool!
We have a winner! No mote calls please.
I asked ChatGPT two basic questions about a commodity market and an adjacency the other day, related to my job, and it gave me an answer to both like a kid who read maybe the headings in last night’s assigned readings and nothing more. The information was outdated, irrelevant, incorrect, and so generic as to be useless. My job is safe for a while longer.
“The price for [food commodity X] has declined dramatically over the last 9 months, why has demand for this commodity not increased?”
“Why did [large group of businesses] all reduce their purchase of [commodity] suddenly in April?”
Everyone wants it to do this. I recall a comment saying it kept finance related information about 6 months old intentionally.
I was screwing around with it yesterday and it said it's updated through september of 2021. It's way behind.
That’s actually good to know and a bit of a relief!
While I suppose that AI could eventually code, I don't think it could ever weld or plumb.
In my mind the concern is less that it really can do my job, and more that it can in the short term make some senior manager think that it can do my job because he has no idea what I really do. The end result for me is the same.
Everything that is written will be read in the abstract.
Just an aside, but stuff like this is why you DON'T want X-ray vision.
You'd be rolling down the street, on your way to the store, casually looking into people's houses and you'd see some naked fat dude in his bedroom fucking a five-cheese burrito.
And your whole day's shot.
You calling me fat?
Hilarious though. Lol funny comment.
Oh yeah, that would be terrible... do you have that address?
I'm starting to understand why space aliens want to invade Earth and wipe us out...
I read the first line and immediately thought of something more hideous, and my whole day was shot.
What's wrong with a five-cheese burrito?
Before or after?
Yikes!
It just seems to be another way for people to get more lazy and stupid. And frankly, there’s already enough of both around.
I’m still astounded that our esteemed VP had some sort of AI summit. The mind boggles. How about asking for a summary of one of her “speeches”? Can a computer commit suicide?
That might happen if you asked for the greatest speeches and pearls of wisdom from AOC! 🤢🤮
You just used "AOC" and "pearls of wisdom" in the same sentence?
Huh.
Thirty years ago my favorite prof showed us the latest version of a 2D electromagnetic modeling package and commented that it did the job that eight guys had done all day, prior to electronic computing. It's a pity that Spiro Agnew wasn't assigned the task of crafting a response to the automation of Finite Element Analysis. The world needed the laughs!
"I predict that in the future, computers will be twice as powerful, 10,000 times as large and only the five richest kings of Europe will be able to afford them."
Yeah, if only.
Interpolation is always going to be more accurate than extrapolation.
The same with the reliability of deductive reasoning over inductive reasoning.
"iambic pentameter"
C'mon Mr. English major. Now you're showing off.
I accomplished something today, though. I know from college speech class that “dyadic” communication is between two people, but I didn’t know what the equivalent is if you’re talking to an inanimate object, like a computer, when you’re instructed to “converse” with a computer on the other end of the phone! It’s “monadic!” 👍
Monads in the Avoidable Contact comments section.
We've seen it all. Time to pack it up and go home.
Thankfully “m,” not “g,” as the first letter!
“An order that can be misunderstood, will be misunderstood.” - Napoleon
"Always remember that it is impossible to speak in such a way that you cannot be misunderstood: there will always be some who misunderstand you.” - Karl Popper
Does Popper's comment apply to LLM? Is there a way, short of using formal programming languages, to make sure that a computer fully understands the input?
I stopped having any real interest in tech once I got bored with my PS2 so this is probably a dumb question.
If AI is nothing more than a super deluxe lefty text generator why is this guy so bent out of shape?
https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/
From writing legal documents to ending humanity seems like kind of a big leap. What am I not understanding?
Yud is all bent out of shape because he has made a career out of getting attention with regards to these matters.
Holy shit that guy is an idiot. Did he just write that we can't be sure GPT systems aren't CONSCIOUS?
A horrifying number of people in positions of power and authority are convinced that these things are conscious or pre-conscious.
The secret to become an executive in corporate America is to be an extroverted midwit or a sociopath
THAT is profound and well-stated.
I'm assuming you've read the locklin peace on AI? Considering I'm pretty sure I only read him cause you linked him
Absolutely!
https://scottlocklin.wordpress.com/2023/03/23/how-to-be-a-technology-charlatan/
Funnily enough Yudkowsky features in the Locklin piece. I feel better now. Back to stroking the prompts for just the right literotica piece.
Not sure if I would’ve used that adverb in that particular context!
That is a brilliant idea.
But legal documents WILL be the death of us.
Thank you for the kind words, Jack; as well for expanding on some of my thoughts in new and horrifying, but ultimately enlightening, directions.
That's a lot of words for "I wish I thought of the Lucy Lawless sex doll."
"That's a lot of words" is going to be my epitaph.
I thought it already was, what with the whole horse thing.
Never forget 🪦
I refrained from calling you Mr. Horse Lover elsewhere but now I see that was a mistake.
Back to Sarah Jessica Parker I see...
(Sigh)
I thought your nom de plume was used in Jack’s “Car And Streeter” stuff at the insurance company which shall not be named!
I thought that writer was driving something nuclear powered, lost control, and hit an elementary school, then a fission reaction spontaneously occurred, and took out the surrounding subdivision! 😬
Oh yeah. 50s Tesla.
That's OBVIOUSLY not how you build an automotive reactor. Everybody knows that.