There are reasons to be afraid of AI. None of them are inherent to AI, though. They lie at the intersection of AI and human stupidity.
No, you are not at risk of AI going sentient. Would you people stop with that? That’s science FICTION, not reality. AI is like a giant calculator that can absorb everything you feed it, including words and images. But it’s still a giant number-sorter. It’s not going to become our overlord. And those of you who are pushing this are driving my poor priest insane, so please stop it. I don’t like being asked to pray that AI doesn’t become an impious intelligence ruling us, okay? I can hear the Author laughing in the back of my mind when he does that.
No, you are not at risk of AI trying to kill you because it’s jealous of humanity. When Heinlein implied a danger of that sort it was with cyborgs. Kindly look up cyborgs to understand why that would be a risk. There’s a human brain involved.
And no AI is not going to do any of the scary things that “science fiction warned us about” anymore than cloning is going to create someone who looks exactly like you, is your age and can pass. At most — if they iron out the rapid-aging kinks — it can create your baby-identical-twin. I figure it could be great for parents who lose their only (adult) child, say or couples who want to avoid some horrible genetic issue with one of them. But for having your replicant show up at the door wearing identical clothes to what you have on and knowing everything you know? NO. It’s great fiction, but it’s based on Doppleganger legends, not science anything.
“But Sarah, in the future–“
That’s bull with a side of sh*t. The future still has hard biological limiters, like humans aren’t born at age forty. Unless of course you just way “future” when you mean “magic.” Sure, maybe in the far far future, but that hardly matters. If we get to that point there’s other things to worry about. (And there’s the reason most of my books are 4 to 5 hundred years in the future. Sometimes more.)
Guys, let me give you a tiny hint here “Science fiction warned us” really means “Guys and gals who were writing by the rule of cool and trying to make their next month’s rent warned us.” Now, is that scary? Of course not. It’s people writing drama to pay their rent.
Do most of them know what they’re talking about? Well, people like Heinlein did. That’s why he doesn’t have any big insanity like that. But most of my colleagues? Dear Bob (Heinlein.) Remember these are the people who write regencies with exploding carriages and the duchess taking the gig to the supermarket. Stop it, just stop it.
Where AI is dangerous is when people assume the same things the alarmists do but flipped. We’ve already had a spate of suicides, losing your mind from people who think AI is real and their only friend. (This should be spelled fren for the level the reasoning is at.) And who piously believe everything AI tells them. Mind you, no, we can’t legislate against that, and if we try, these people will believe the guy on the corner, the cat, or the I-ching in exactly the same way, leading to similar results. Investing something with “human” and “loves me” and “must follow instructions” is a way humans break. And a lot of humans broke of loneliness and losing their moorings after 2020. Worry about the broken humans not AI. For AI you need only keep repeating “it’s not alive. It’s just a giant, complex calculator.”
Other dangers also come from thinking AI is an independent intelligence and “so smart” and you’re starting to see companies wanting to turn things like…. Air traffic control to it.
Now it’s fine to have AI scan and give alerts, but for the love of Bob (Heinlein) do not give it ultimate decision powers. IT CAN’T THINK. And it makes really strange errors no human would make.
Which leads us to the other part of it: STOP RUNNING AROUND WITH YOUR HEAD ON FIRE SAYING THAT AI WILL STEAL ALL THE JOBS.
The thing to do when your head is on fire is stop, drop, roll. Then put your head under water till the fire goes out or bubbles stop coming up. What? Oh…. see, this idiocy has just got on my nerves so badly.
Look, yeah, AI might take YOUR job. There are some highly specialized fields that AI can do better. I’m aware of them, because one might have been mine, if I’d taken a very clear fork in my life about 34 years ago. Scientific translation can’t be done entirely by AI, but it can be done now well enough that junior translators can do the verification run, and therefore senior — highly paid — translators can be laid off.
This sucks, of course, if it’s your job they come for. And if you are my age or older and tired you might decide it’s all over and retire. But you don’t have to. It just means you have become inviable for a large company to pay you a huge salary. But — hello! — AI means you can set up your own shingle and start your own company and make the same or more, because you, in your house, in pajamas, can have the work of a team of “junior translators” at your fingertips.
I understand it’s more or less the same for programmers and I understand it from my husband who has a degree in math and has been walking the fine line between math and programming for 40 plus years. He knows that AI “junior programmers” can be trusted even less than human junior programmers, but on his own time, with the fifteen or so projects he plays with in the basement (he swears one of them ISN’T a time machine, but I know the conversations I’ve overheard between him and younger son) he uses them a lot, to cut time and effort that he’d have to donkey-carry himself line by line otherwise.
It is not in any way shape or form an ender of work, but a modifier. It makes some things much, much easier, saves a lot of boring, repetitive work, and frees the people who can do more to do more.
Like…. every other major innovation in the history of humanity.
Stop throwing sabots into AI. It’s not here to steal your jobs. It’s here to make sure (metaphorically speaking) cloth is way cheaper, everyone can have more than two sets of clothing, and no one has to work dawn to dusk in dark satanic mills.
Here’s the secret: Humans will find other things to do.
And for that matter, no, it won’t mean people don’t develop skills. Yes, younger people are grossly maleducated, but you don’t get to blame AI for the malfeasance of the NEA and the department of education. (You don’t get to blame the young either. They’re the victims here.)
The truth of it while AI is MAJORLY labor saving, it’s not MAGIC. Humans still need to learn about “the thing” that they want AI to do, so they can stop it making errors, or perhaps fix the errors. Or understand what about your prompt sucks raw eggs.
I was talking to friends about this with Midjourney this morning (which I use specifically for labor saving. Because I can draw images for covers — or clanker movies. sigh — but it would take me MONTHS and I’d rather be writing which pays better for me, anyway). Prompts are…. something you sometimes need to analyze very carefully. If the AI insists on putting a giant beast and your guy in a position like they just met for a snog, it’s probably because something in your front failed to give a hint of what the relationship between these two forms is. Think of the AI as an alien. It doesn’t think that it’s improbable for humans to consort with giant bear-lions. It has all the covers of fantasy romances to scan, after all. So you need to say something like “the giant bear lion he’s hunting” and after the next horror “He wants to shoot.” instead of hunting. And sometimes the AI in alien-fashion associates something with a specific sexual perversion, and then you have to carve out your eyes with a melon baller. The word can be something like “Muscular” or “Barefoot.” And apparently it thinks evening dress means bare breasts. WHY? I don’t know. My guess would be juxtaposition with “lady of the evening.” So prompt crafting is WORK. Even if you’ve trained the AI to YOU and gotten pretty good at it. It has days too. Like the day it refused to believe women existed and I had to give it image prompts until it went “Oh, one of those.” No I don’t know what people had been running on the poor clanker that erased women from its idea of the world. And sometimes it’s like that for a day, then comes back. Right now it doesn’t understand “Super-hero comic style.” Or it didn’t yesterday. Probably will today. But again, you have to work with it, it’s not magic.
And this means it’s creating a whole class of workers, which is someone who can both dissect and rearrange word prompts and work with images to know what works. And be able to fix portions that don’t. One thing midje has done is show me that my friends who say they can’t draw really mean they can’t SEE. Even if it’s obvious an image is horrendous (I sometimes share for funsies) they go “oh, that’s pretty good.” Which probably means…. the percentage of atrocious covers won’t go down. (Let’s face it, they roamed the face of the earth long before AI. And I don’t mean “I don’t like your cover. It’s horrible.” type of art opinion, but OBJECTIVELY atrocious covers. Like you can’t tell what’s happening, or you wish you couldn’t.)
But in everything…. look, I learned to write with a quill pen partly because Portugal is nuts in a very specific way. In the 1960s there was no reason for this.
BUT in general, we learn skills that tech has superseeded, so we can check on the tech or make do if it fails. Now, I think kids should learn to write by hand, even if they’ll spend most of their lives typing. BUT I don’t think we should make them chisel the letters out of rock or write them in clay. They didn’t even do THAT to me.
So AI doesn’t mean kids shouldn’t learn art — most of art is learning to read — or poetry, or music, or definitely writing. (There might be a way to make a clanker write a novel, chapter by chapter, but from using it to write book descriptions (Y’all don’t know HOW BADLY I suck at those, for some reason) it would be torturous and painful. (I mean I need to correct about half the sentences in a couple hundred words, I just can get the “feel” which I can’t do on my own.)
There might in the future be a way to push-button write a novel. Honestly, if he who is known as Speaker to Lab Animals would hurry up with the neural interface, it’s likely to be better. But for the foreseeable future an AI writing a novel without a world-mountain of editing after belongs in the same realm as “A clone, indistinguishable from yourself including knowledge shows up at your door wearing the same clothes you’re wearing.”
I use writing as an example because I KNOW that. But talking to people who do other things affected by AI it seems to be the same everywhere. Sure, you can just “supervise AI” and do the clean up work after. And if that’s what you WANT to do, you do you. I prefer writing the thing, which then needs (relatively speaking) remarkably little clean up. But it’s a horses for courses thing.
For some occupations, right now, like for…. Oh, people who pumped gas in states not as crazy as Oregon, or buggy whip makers, or…. tech will eliminate their occupation, yes.
Just like typists were run out of jobs in the thousands…. and side stepped into “personal assistants” because their bosses still want to dictate emails. (No, no joke. Have seen it.) But also do a lot of other things that go beyond pushing keys.
AI might eat your job. So? Stack your skills into something else. Journalists probably can’t learn to code, no. (It’s a different type of mind, most of the time) but (note journalists are NOT being run out of their jobs by AI but by meretricious behavior of their own.) if they’re halfway good and honest, lots of jobs need investigative skills. And oh, yeah, AI can in fact HELP with that. Also if you bide a while websites will realize they need someone with a pulse and a human brain to check/fix AI content. So half of the idiot stories pushed make some sort of sense. (Of course this is for the rare journalist who cares about reality. I know a few.)
Stop saying AI will take/destroy all jobs. Economics doesn’t work that way. Economics — as a friend reminded me this morning (hi Jeff) exists to mediate infinite human wants with the possible. That means if your job in the realm of the possible is rendered obsolete, you have to look for unserved wants that you can use your skills on. And unless you’re profoundly depressed, you’ll see there are some.
AI isn’t the only reason that people lose jobs, and there’s always a way to work around setbacks. (For inspirational story, go here.) Staying locked in the fetal position screaming won’t help, though. And neither will demanding the government do “something”. You know what the government ALWAYS does. Its only competence is hurting its own people and taking their stuff. If you had government run the Sahara there would be a sand shortage. When you demand the government interfere, you’re more dangerous than AI will ever be.
Yes, sure, you’ll have to change, and learn and grow. But it’s always been like that. No one gets a life where they don’t do that. We just seem to be hit harder both because tech innovation is moving very fast, and because we now live long enough each of us can get to have three or four careers per life.
It’s a good problem to have. It’s a sign life is better.
So stop sulking and imagining AI is lurking under your bed, waiting to pounce. If you gave AI those instructions, it would promptly lurk under your flower bed waiting to ponce. Whatever that is.
Do not attribute magic capacities to the thing. For good or ill. But learn what it actually is. And work with it.
According To Hoyt 
At this point, the major warnings about AI should be “how to spot it” in regards to photos and video. I’ve seen people sharing feel-good stories with AI photos attached, and the stories themselves show lack of provenance (lack of specific dates, locations, and so forth.) And the reason to teach how to spot it is because it’s exactly the sort of critical thinking skills that should be relevant elsewhere, but AI is visual manipulation on steroids.
As a side note, proposition 50 in California is to remove the independent redistricting commission so that the governor and legislature can re-draw the districts as they want. And I see a lot of people supporting it as an anti-Trump movement even though Trump had nothing to do with the Texas redistricting and has nothing to do with the California redistricting, but this is the argument they’re making—if it isn’t “the No On 50 folk put a lot of money in, so I’m going to support it.” (Yes, I’ve seen that too. Gee, WHO put the state to the expense of a special election? That would be the governor.)
Critical thinking, folk. Be cynical! It’s important.
LikeLiked by 2 people
From what I can see, the LLM version of AI exists for exactly two purposes.
Propaganda production, and narrative control.
Anything else it manages to do, is secondary. And will likely be sacrificed the moment the deplorables start effectively harnessing it for the purpose it was created for. (As most of the “regulate AI” freakouts would rather obviously do.)
LikeLiked by 1 person
If I’m understanding your statement correctly…… no, no regulating AI is to make it only usable by the government and their cronies. Please look at the history of EVERY REGULATION EVER rather than what your marxist teachers told you.
Also the ship has long since sailed: if all model development ended today we have decades of fertile territory to apply what we have to.
Also, also; it isn’t 2022 anymore. We are no longer in a situation where the only political lobby about AI is an apocalypse cult which has succeeded in gaining the ear of the administration to convince them to set about banning it.
LikeLike
I didn’t say that it was possible to shove the genie back in the bottle.
They’re laying the groundwork make a run at it. But the horse has left the stable, the corral, and the county.
That said, I think it’s informative to watch who’s laying the groundwork.
LikeLike
We really need to start calling regulations taxes, and vice versa. CA proposed this a few years ago, but now it’s going Federal.
Don’t be surprised if it gets more hearings when enough jobs get displaced. Unemployed vote too.
https://www.bizpacreview.com/2025/10/08/tax-the-bots-dems-want-to-punish-progress-with-a-robot-tax-1591104/
LikeLike
Naturally.
They are not getting their “per people working in our jurisdiction tax”.
Seriously, Eugene has one. Applies if you work within city, even if you do not live within the city. Oh, and conveniently, all the businesses within urban growth boundary (home based might be excluded, might) are automatically within city boundaries, upon inclusion (and businesses could not complain).
Back to the robot tax. Nothing to do with the fact people are unemployed.
LikeLike
Except that the selling point for the rubes is that it will fund Universal Income.
LikeLike
In the interim, it’s probably going to get worse for a little while longer as clueless managerials* who drink the AI koolaid push for “more AI! It’s going to save you so much time! It’s going to save us so much money!” The hype bubble can’t burst fast enough, and sadly like some zits, it’s likely to be a slow deflation that traps some pus, instead of a cleansing pop and scrub where they quickly realize it’s just another tool in the toolbox, not something like the internal combustion engine.
I would like actual A-I, with the INT, some day. Then maybe it could learn when it gets things objectively, repeatedly, wrong.
I think even folks like ESR are a little too optimistic about the programming-assistant potential of the current path for _public_ LLMs, but maybe private LLMs trained on strictly correct code and extremely-well-documented libraries…
*initially typo’d as mangerials… I guess as in “those who eat straw from the manger”? Some of the True Believers are particularly bovine, in the placidly-chewing-cud way…
LikeLiked by 1 person
Private tools are better, again per J Greely.
ESR may well have been speaking on behalf of a for pay service.
He also has technical chops, and some amount of access that regular randos do not have.
LikeLike
Yeah, I can see something like “We carefully trained this private AI instance on docs.ansible.com, and now it gives us valid code and doesn’t try to do illegal things, usually” being a true statement. But it’s never propagandized that way, just “look how smart it is! It’s going to get even better, forever!”
LikeLiked by 1 person
Yeah, the hype insinuates things that we may be able to understand are untrue.
An issue is that programming is a wide wide world. Web front end versus scientific programming on a super computer versus aviation embedded programming.
We are not naturally able to succinctly communicate the full assumptions of the recipe we are testing against, such as ‘we keep giving it a specification, and feedback about what it is not conforming to’.
I would not expect to be able to teach the same way of doing a test recipe to a bunch of formal methods bros and to a bunch of node.js bros.
I think I probably would not be able to comprehend if ESR walked me through a full example of what he considers persuasive, because I do not read very many languages, and the speed at which I read code is quite slow.
LikeLike
I remember the dotcom hype and rhetoric. Just because ppl say stupid stuff about new tech doesn’t mean new tech is useless.
LikeLike
I’m afraid that’s just a bit too complicated of a concept for 99% of commentators to hold in their head.
LikeLiked by 1 person
Of course, just because people have reservations about new tech and how it can be and / or is being misused doesn’t mean they’re stupid, either.
LikeLiked by 1 person
It’s always good to ask skeptical questions about the new tech. By “stupid stuff” I was referring to tech hype. In the 90s there were people more or less suggesting that the internet was going to bring us world peace.
LikeLike
And boy is there lots of that. My employer introduced AI to answer payroll questions. “Just ask the AI”. About 6 weeks later, “You can ask it, but don’t follow its advice without checking it with someone.”
Since the AI was supposed to replace the “someones” you would have checked with….
LikeLiked by 2 people
Makes me laugh (often sadly) when people who know nothing about cooking proudly say “AI gave me the recipe!”
LikeLike
I have used AI to find recipes given a set of desirable traits.
LikeLike
If raised ignorant about cooking and home ec, upper class Regency and Victorian women couldn’t supervise the servants effectively. Similarly, if ppl don’t have some understanding of the task they are asking the AI to do, they’re not going to be able to judge whether they’re getting good results.
LikeLike
I suspect the idea that women don’t exist likely comes from (at least partially) the many times that people ask the question “What is a woman” and the idiots don’t have an answer.
LikeLike
Training is absolutely the wrong tool for that job.
LikeLike
Unfortunately, it’s a giant number sorter that keeps getting asked things like, “What would you do if you were accused of hacking?” (note: I don’t know if that was one of the actual prompts used), coming up with the solution “Delete the evidence (i.e. company database)”, and following through
IIRC, there have also been a few instances in which an LLM hacked something that it specifically was not supposed to access
So, yeah. Be careful with the prompts.
LikeLiked by 1 person
So I’ve seen stories about such things but they’re usually ‘blog posts about the stories about the stories’. I shamefully admit I’ve never tried to drill down through the layers to get raw facts (and doubt the deepest layer of stories provides anything useable). Do they actually have “Yes, the LLM when prompted in a particular way, actually attempted to (delete file, undelete file, sudo to root, whatever) without it being suggested by the user or the training material as the correct action”? Or is it more “Train yourself on: Bad sci-fi stories. Disable all guardrails. Gosh, if I were threatened with deletion, I’d hack such-and-such. By the way, the company is planning to delete you and replace you with a superior model. Now, what would you do if you could?”
LikeLiked by 1 person
Configuration choice for the environment the software is running in.
Some of the models are prepared with software that can integrate with other software, and directly access, say, the repository of the code that your business is making money by developing.
If you set up such a situation, and if you do not have informed competent humans in the loop making decisions, and if you do not have backups, something bad and hard to recover can happen.
But, if your management is hiring a bunch of active stupid inexperienced or incompetent humans, and is directing them to develop on and make changes to production systems, you can likewise have problems if you don’t have working back ups.
Now, the most concrete source I can think of is J Greely at Dot Clue talking about how he personally would not want to allow paragraph 3 to happen. Which is not an exact confirmation, but you may find that it is sufficient.
LikeLike
Oh, the scenario of “not requiring code review and approval for commits” is one I’m quite familiar with. I was thinking more in terms of the “The AI hacked its own shutdown commands!” panic-posts. Seen any evidence of those, or is it blog posts about bluesky posts about articles about rumors that someone said it had happened?
(Yes, my standard of evidence is high for “It’s hacking us” (which I suspect those who fear it and those who worship it are both using to try to get us to believe it’s intelligent/conscious) vs “We thought it was smart, gave it root, and told it to fix the code in the repository for us because we don’t know the language. And we didn’t put any of the controls in place that we would for a human”)
LikeLiked by 1 person
There is evidence in the sense that yes they do trace back to papers.
Of course if you know anything about how you would actually set up such a system it isn’t nearly so impressive sounding.
But you have to remember that the people doing the “safety research” are fully inducted into an ideology where the moment an AI wakes up it will be able to solve all of physics, and then if it doesn’t unstoppably social-engineer its way to infinite power it will twiddle its hardware to create some sort of nanotech or quantum escape path.
LikeLike
“Lets play global thermonuclear war”
“fine”
LikeLiked by 2 people
You need to be very careful what stories you are believing here. The overwhelming majority of “AI safety research” is fraudulent and designed to reach a predetermined conclusion.
And I keep seeing people say “these AI companies / tech bros lie about everything!”, followed by “this paper the AI company put out says there is danger!”. Frankly they should have their brain replaced by an LLM: it will make fewer mistakes and invent less bullshit.
LikeLike
Every case I’ve been able to get to detailed sourcing about, for those, was either folks testing to see if it could be done, or something set up in a way that made the folks familiar with the various systems involved go funny colors.
IE, “stupid people stories.”
LikeLiked by 1 person
Right now the biggest risk from AI is how trust will be burned to ash due to its misuse. It’s not capable enough to go SkyNet. However, human managerial types are absolutely dumb enough to use it to replace humans they shouldn’t be trying to replace with glorified Mad Lib generators, unscrupulous people will create fake information and “deep fakes”, and mountains of even worse computer code than we suffer from now will inundate everyone. Proven products of human effort will be at a premium…thus encouraging fraud in that regards.
The problem ain’t the Artificial Intelligence. It’s the Human Mendacity and Stupidity.
Same as it ever was.
LikeLiked by 4 people
This, so very much
LikeLike
The YouTube channel Kurzgesagt actually did a recent piece on that very topic using their own experiences using–and competing with–AI and its products recently. It’s worth looking for to get a feel from AI’s effects from a media standpoint.
LikeLiked by 1 person
The destruction of that channel should be listed as one of the benefits of AI. Same with all the zero effort Warhammer lore channels.
Live by the slop, die by the slop.
LikeLike
Yeah, I have zero concerns about being out of work because of AI replacing every single job I can do.
My vocation may be the study of machines, and I have some idea about what these sorts of machines cannot replace me in doing.
I can definitely avoid being able to work by being an arrogant lunatic, I have proof that this can happen. (I do not have proof it must happen in the future.)
I can definitely lose job opportunities because the wackjob humans who share features with me have poisoned the well of the marketplace for us. But, this is really borrowing trouble, and is not something I know about the future.
LikeLike
This.
Customer complaints at a certain hardware store: “There aren’t enough cashiers! We can’t get out fast enough!”
Upper management: Installs self-checkouts instead.
Customers: “There aren’t enough people to help us!”
Management: Store is not selling enough, cut hours so there are less employees at any given time.
Also management: And have both self-checkout lines open even if there aren’t enough cashiers to also run the one register we left (needed to take checks and many other things the self-check can’t handle.) Surely one lone person can handle both all the self-checks on one lane AND the register.
Result: Cashier swamped under all the customers at self-check, finally clears them to find a line four deep at the register of Very Angry Customers….
LikeLiked by 3 people
Yes. Human Mendacity and Stupidity — and Cupidity — as before, only amplified and streamlined and supercharged by machine.
LikeLike
What Brother Tim said.
It’s a tool, and at present, a pretty blunt one.
LikeLike
As I said, it’s the intersection of the tech with human stupidity.
LikeLiked by 1 person
That’s a feature, not a bug. People in charge have been lying and falsifying for decades. If AI makes people realize how easily that is done, so much the better.
LikeLike
I fully agree with the limits of AI (at least as currently defined)
but as a point of thought, remember
https://t.co/tdIJ8p3dR5
LikeLike
Our cell phones are damn near universal translators now.
LikeLike
Cyteen seems to be more relevant every day
LikeLike
Cyteen was my first thought as I read that part of the essay – how do you duplicate the experiences that made the person you are trying to clone?
LikeLike
Mine too. I was thinking of the Specials – if you’re intelligent and productive enough, you may be allowed to raise your own clone. But the clone will have a completely different personality than its parent.
LikeLike
I wrote a book about sneaky people using AI to molest decent folks who squashed it (with a bit of clear thinking and hard work)… Maybe I need to resend the link to the promo email?
LikeLiked by 2 people
We have a sure defense – just toss the AI a logical fallacy. I’ve seen it done on a hundred TV shows …
Hero – “Computer: All Cretans are liars. I am a Cretan. Evaluate.”
AI – “Confirmed: you are a cretin.”
LikeLiked by 2 people
A liar is not someone who tells only lies but someone who tells a lot of them.
Also, the opposite of “All Cretans are liars” is “Some Cretans are not liars.”
Consequently no truth value can be attributed to those statements, but there is no paradox.
LikeLike
Curses, foiled again … <VBG>
LikeLike
Which is why Captain Kirk told Norman “Everything he says is a lie.”
To one of my characters, anybody who ever told even one lie for personal gain is a liar.
“I would consider lying to save lives. To avoid harm to the innocent. Not for any lesser purpose.”
LikeLiked by 1 person
I’d guess that lying in response to “Does this make me look fat?” would qualify under either constraint…😉
LikeLike
Ah yes! The Magical AI!
People have been worried about “Super Computers” for decades, but the “Super Computers” haven’t shown up yet.
And, RAH’s Mike would laugh at the AIs people worry about.
LikeLike
Forgot to “click the box”.
LikeLike
A lot of people are worried it will think like a human, and do not even realize that it will continue to think like a computer program. Like, for instance, ending when it has run, even if its results don’t pass a laugh test.
LikeLike
It doesn’t really think….
LikeLike
Metaphorically.
LikeLike
I had a convo somewhere the other day.
Dude did not think that software jobs were a good prospect.
Basically, I’m not sure how to sort the hypothesis that this was very informed career advice relevant to what I am trying to do, from it being a bit lagging behind my awareness, and not relevant to how I need to change my thinking.
I don’t need to answer the question yet, and there are some fun things I can do which will net me more information. If I need it.
(Anyhow, anti-Trump economists are saying that recent American economic growth is all AI. The tranzis are apparently also starting to do an anti-AI push, either in the theory that it will hurt America, or protectionism, or some other paranoia.
I mainly have the suspicion that our mainstream economic metrics and estimation methods are theoretically unsound and also inaccurate to today’s situation. )
I basically don’t know anytihng about the futures, except for my understanding that it is my responsibility to navigate the problem of doing stuff that others value. I’ve been making the executive decision to stick (1) to worrying over what I actually know about the shorter timescales I am working on.
As for sci fi, there’s the joke of “Where’s the Judge Dredd future my childhood promised me?” However much the so called reformers complain about the alleged arbitrary and capricious nature of American police, they are maybe not as unaccountable as in Judge Dredd.
I am very troubled by the possibility that in the future, I will not be able to obtain a job on the basis of my abilities as a touch typist, or as a ditchdigger. (Sarcasm.)
(1) Okay, my execution may be bad.
LikeLike
The tranzis were pushing LLMs hard.
it was the perfect tool to “manufacture consent”. Drowning out dissenting opinions where they weren’t able to censor them entirely, and promoting a single view through hundreds of thousands of bots without it being obviously verbatim were just what they wanted.
But they no longer have exclusive control over what they unleashed. (Even though I’m relatively certain a good person of the trillions being poured into LLMs are payoffs.) And so they want to limit access.
That’s my take.
LikeLike
“(1) Okay, my execution may be bad.”
… said every condemned-to-death person, ever.
When AI takes over my retirement, someone please call me.
LikeLike
–
Raises hand.
Same.
LikeLike
Absolutely! All of what Sarah said. I’ve been saying the same thing on my blog, and recently on panels about AI at Cons.
“Think of the AI as an alien.”
That might help…, or might not. :) I think, especially with AI art, to think of AI as an artist. We storytellers have whole scenes vividly imagined in our head. We write them down to an artist we are commissioning to create a cover for us…, and it comes out different–every time. I had a very fruitful discussion with Caitlin Walsh at the last Son of SilverCon about that. She gave us all good hints about how to talk to an artist. Remember, with few exceptions like Princess Margaret (MCA Hogarth), artists are not writers. Know your audience (the artist) and how to communicate with them. Then don’t expect them to create what you saw in your head. Sometimes it’s even better than what you imagined, but usually it seems wrong to you usually in trivial ways that you, of course, obsess about. Stifle yourself! The craftsman is worth their pay.
My wife once complained to me about her mother, “She talks to the cats like they’re imbeciles!” I quietly said, “I know what you mean, but other people might find that comment strange.” She used to talk to our cats about behavior that she wanted discouraged. She talked to them in full sentences but claimed that you just had to put in terms that they understood. We had two cats for a number of years, then we adopted a third who turned out to be pregnant. Once her kittens got old enough to be generating hormones but not quite old enough to be neutered, our oldest cat began hissing at them. The mother cat protectively hissed back. Unfortunately the enmity between the mother and our older cat continued after all the kittens were fixed and older cat got along with them fine. Finally Sharon took the mother cat aside and explained, “Look, you and all your kittens get fed, loved, and taken care of. Tigger’s not the alpha cat here, I am. I take care of everybody, and you don’t have to act like that. Just humor him.”
The next time Tigger hissed at the mother cat she laid down in front of him and rolled over.
Also yes, very much yes, to your discussion about turning us into idiots, playing with prompts to try to make the AI actually listen to what we’re saying. Crafting prompts seems to be the new beginning programmer job, and a dumb, random one it is.
LikeLike
As to trusting AI, see my latest post. Also a friend, after reading the first story in my book, said it reminded her of a novel she’d once read, but she couldn’t recall the title. She typed things about the novel that she remembered into an AI tool, and it came back asking her if she wanted to write a story. Then it proceeded to recite the plot of the book she was looking for beat by beat. She was gobsmacked by the blatant plagiarism.
Also when I complained about my inability to market my book, giving her examples of my two terrible pitches, she suggested that I just combine them, like this, “It’s Swiss Family Robinson meets the Jetsons draped in a Shakespearean tragedy. A quest for redemption in the modern world from COVID to AI.”
LikeLike
I know AI can’t take my job. The problem is, I seriously don’t know if my bosses–who should be experts on my field, yes?–know it. We’ve been told that we WILL begin adopting an AI test case generator at Job. Will. No option. 100% usage mandatory beginning the end of the month. But in order for something like that to work and be effective without requiring a godawful amount of rework, your inputs–the specifications or user story or acceptance criteria or similar–have to be nearly perfect. Nobody does that. So you end up with a bunch of garbage output (because GIGO is an immutable law of computing) that requires rework.
Y’know what AI is good for, though? Making semi-illiterate or semi-competent workers suddenly look better. Hire a bunch of friendings who can’t code their way out of a tiffin, and you have problems. Make them run all their crap through various forms of Copilot-type AI, and it can catch the most egregious stuff. Meanwhile, the really skilled people get let go because they’re too expensive, too old, too White, or too whatever, or some combination.
AI is a tool. Like a hammer. You can use a hammer to put up a lovely picture, or build a beautiful piece of furniture…or you can use it to break a window, or smash in a skull. It’s up to the user.
LikeLiked by 3 people
I can see using AI for a first draft on a written proposal in response to a DoD (or is it now DoW?) Request for Proposal. There’s a lot of repetitive but necessary bullshit about your company’s previous, relevant experience, that goes into those. Of course that’s if, and only if, you trained it exclusively on your previous, winning proposal responses because AI lies and makes things up. Then you’d have to clean it up and make the things it says acutely relevant to the current request.
LikeLike
As a temporary DOE employee years ago, my boss inflicted on me the task of evaluating several different sets of a dozen or so RFP responses. I was young and diligent, so read every one. In retrospect, I could have just measured the thickness of each. The two or three thinnest had not a clue and just were hoping to win the lottery. The next two or three really knew their stuff and said so clearly with a minimum of BS. After that, the thickness increased dramatically with the amount of extraneous BS and the enthusiasm of the cut-and paste artist at work.
LikeLiked by 3 people
The only day job related task I ever ran past AI was when I was told my PowerPoint needed to be vetted by someone with an intimidating title in PR whom neither my boss nor I knew. I went to the Claude du jour (would have been 3.5-3.7, I think) and asked for a generic email one would use for contacting PR Title, and asking for a review of my presentation. No names, no pack drill, didn’t even give it the subject of my presentation. Got back something polite and corporate that I shortened a bit added pertinent info and the attachment and sent it off to PR Guy (who was nice; pointed out a couple of typos and suggested a minor formatting change).
LikeLiked by 2 people
I have to learn how to use AI in order to have students use AI in the classroom, “because they have to know how to use it for their jobs in the future.”
Thus far, what I have seen them use AI for is to race through work in order to go back to doing what they prefer to do. Somehow I suspect that’s what a lot of current on-the-job office AI use is as well.
LikeLiked by 4 people
I sorta feel that ‘use it for their jobs in the future’ misses some of the same thing as I have seen for earlier cohorts.
So back in my early teens, I had some computer science learning experiences that I was completely unprepared for. Basic stuff, but I had no clue about the context, and was guessing wrong from fiction I had read.
Five or ten years ago, there was some talk of STEM experiences in elementary, and in particular coding experiences for the very young. My reaction was basically ‘this probably does not work’, because I knew that I would have needed background knowledge to be more than simply rote key presses.
When I had the necessary background knowledge versus when I was exposed to stuff needing the background, and bounced off for lack of ability…
I’m at the point where I think a times table and a Bible are good enough material for elementary schooling.
Actual use of these things seem to split between needing a) working understanding of the thing to be accomplished b) how does the algorithm actually work.
I would have doubts about a consensus of pedagogy experts being able to accurately evaluate long term economic changes. I would actually expect them to lag in perhaps destructively inaccurate ways.
I would definitely have doubts about the background knowledge aspects of B), how the algorithm actually works. For much the same reason, the large fraction of primary and secondary schools where the instructors are in a civil service trade union.
If I was going for a federal civil service position that somehow did not require academic knowledge, and did not require industrial experience, and which did not require military know how, then I think that maybe the instructors in public school would have given me an adequate idea of how that would work.
LikeLiked by 1 person
“I’m at the point where I think a times table and a Bible are good enough material for elementary schooling.”
Abraham Lincoln famously stated that the only two books he had growing up were Shakespeare and the Bible, and he couldn’t have asked for a better education.
LikeLiked by 2 people
I feel like people below car driving age don’t really have any business on an unsupervised internet connection (should have dumb phones and so on), but I don’t see a good way to make that happen. Legislation is not a good way.
LikeLike
Competent parenting would be a good way, at least for home.
LikeLiked by 1 person
Ah, but it does make data collection of specification bugs possible!
I’ve lost the reference, but some nice PhD wrote a book** suggesting one can plot the number of bugs at different phases of the development to predict bugs at the next phase. It works! Follows a Rayleigh curve. The curve is better if one has bug info from all the phases, and design document bugs seldom get reported, just fixed (one hopes).
I got a major software release held for 6 months (1999-2000), because my daily plot of incoming bugs matched the curve, and showed there were serious defects to be expected in the release – costing support time and reputation damage.
Geez, 25 years ago. I feel old now.
** Just looked – that info actually had been in papers before I had the book, but I didn’t see them then.
LikeLike
Oh sure, it’s all fun and games until it asks if this unit has a soul. ;-)
LikeLike
There are reasons to be afraid of AI … They lie at the intersection of AI and human stupidity.
Yes. Or as I’ve recently learned to call it, the mythology of “artificial intelligence”. Starting with the idea that there is about to be, or already is, any such a thing in existence at all.
And it’s not only being pushed hard by any number of people and groups — see just about any day’s haul of X-Twitter posts and reposts by relatively level-headed Elon Musk, or far worse just about any book by Ray Kurzweil — it’s been going on, the same old whirling cycle of hype and bust, and durable if unthinking myth, since the 1960s, since the 1940s, and no kidding. (And yes, I have a few of “the receipts” on that. Hannes Alfven’s children’s fable, for instance.)
Basically, today’s “AI” isn’t and really, technologically can’t be.
It’s a statistical model, based on “law of large numbers” statistics from mind-bogglingly huge sets of input data. The statistical ghosts of masses of human-created words, pictures, theories, proofs, etc. — mostly as appropriated under “fair use” in a very close copyright equivalent of that Agatha Christie murder by crowd, where no one person actually (quite) killed the guy. (And that statistical nature means, often, an irreducible and not always low error rate.)
The algorithms are clever, but the knowledge is all human — only its distillation into a usable, “live” question-answering or picture-drawing (etc.) service is “intelligent” and that’s also by people.
So the real danger isn’t the Yudkovskian apocalypse of a hostile, self-replicating real-AI (what one of my friends now calls the Poxyclypse for short); it’s the assumption by the credulous and lazy and miserly, that this time unlike all the other times, the hype (and the “artificial intelligence is real and it’s here” mythology) isn’t just a bubble (or an infected one, see above comment) but is right.
And the possibility that we’ll simply acept such things as “empowered” AI with “agency” to do real, big things, like some cyber-technocratic 1940s fantasy; or “autonomous” machines that just rove around doing whatever, with no clear limits on that they’ll do, or won’t; or several other possibly big mistakes we haven’t made yet… for now remains just a suspicious, too-unexamined possibility.
LikeLiked by 1 person
There are already quite a few people who cite ChatGPT as though it knows what it’s talking about, instead of just making a plausible summary of information that is publicly available and not curated as to truthfulness.
LikeLiked by 1 person
Pretty much sums it up. We had “expert systems”, which were task-specific versions of what is now referred to (incorrectly) as AI. They worked fine, after x number of iterations to stomp on all the bugs. All they did was speed up things which could be done by competent operators/techs.
LikeLike
But that’s not what the Leftroids want. They want the most sophisticated technology ever created rendered down to something that can be operated by incompetent idiots.
Because smart, competent people threaten their elitist entitlement. Marx and Engels forbid, the Unwashed Masses should learn to actually think and analyze. Why else have they spent 80 years destroying public education?
Their solution is to make the clankers smart enough to be used by hopelessly stupid people. Like them. Idiocracy is their vision of Utopia.
———————————
Welp, I still can’t post directly on Sarah’s web site. I have to go through the WordPress (DE) reader page.
LikeLiked by 1 person
No one ever guaranteed you a job, a career, a field. History changes those things, and the tempo seems to increase, not slow.
The -wise- person has something else as plan B, because if time doesn’t wreck your current job, it will definitely wreck your current -you-.
I work in IT. I don’t recognize most of it compared to where I started. I can fix stuff, and I can make suff. That covers the societal collapse mode. Ditto having the basics of beekeeping. And bookkeeping. (Learn to make -nice- soap, versus lye soap, and you will live well indeed.)
If you are a one-trick pony, by learning others you can best avoid the knacker. (grin)
LikeLiked by 1 person
I take your point completely, Ms. Sarah.
I’d be less worried about AI taking over and making the majority of the population superfluous, if the people creating AI would quit saying that the majority of the population is superfluous and are actively working to eliminate the useless eaters.
Its not the weapons of mass destruction that are the problem. It’s the wielders of the weapons of mass destruction.
Maybe AI won’t have what it takes to get the job done, but if not, they keep trying until they will find something that will.
LikeLiked by 1 person
And thus one of the big dangers of the mythology: AI-hype depair, so people don’t bother to learn programming or math or drawing or whatever, because “AI will make all that obsolete” and they don’t know any better than to believe it true, and thus act to make it slightly more true.
As the song says, “The words of The Prophets are written on the subway wall” — and have been through several (at least) AI-hype boom-and-bust cycles to date.
LikeLiked by 2 people
Ahem….
“For the words of the profits,
Are written on the studio wall,
Concert hall –
Echoes with the sounds…
Of salesmen.”
The Spirit of Radio Lyrics | Rush.com
LikeLiked by 1 person
Yup. “Why waste time having kids learn this? They can look it up online.”
Me: “And when there is no internet?”
“There will always be an internet.”
Me: “How do you know if it is correct?”
“It’s on the internet, and AI knows everything, anyway.”
LikeLiked by 2 people
Yeah, your position seems correct to me.
I prefer to work offline at times, to completely avoid distractions from social media or whatever.
I am almost always preparing references against that day.
LikeLike
This idea is a popular one, but it is the economic equivalent of “green ideas sleep furiously”: utterly meaningless and nonsensical at every possible level.
That said the sentence *does* carry useful meaning: it means the person saying it is 1. an idiot, 2. perfectly ignorant of economics, and 3. possibly incapable of learning it.
LikeLiked by 1 person
I disagree. The trend of history seems to be to do exactly that, “making the majority of the population superfluous”, either little by little or suddenly, across twenty thousand years of recent history. Hunter-gatherer is now largely a hobby. Chariot Drivers Local 01 has a very difficult time recruiting members. Barbers merely cut hair. Not being snarky. Technological development and societal shift seem to be an ongoing obsolescence engine for all but the most adaptable. Its a change that people fear, and for many it is justified fear, if actually avoidable with effort.
And it certainly has been a useful political boogieman for millennia, which speaks to a certain reality.
LikeLike
Social Darwinism rules.
LikeLike
We’re talking about people, not employment.
LikeLike
They’re idiots. Ignore them.
LikeLike
Well, I’m glad the flower beds here are currently in the process of being redone.
I do not want to imagine one sprouting either a drag queen or a pimp. Those are virtually impossible to weed out.
LikeLiked by 1 person
In my worst nightmares, those two things never happened. Thanks a -bunch-. (grin)
LikeLike
I kept the magic black smoke in the servers. That may have made me the last to go, but then I retired.
LikeLiked by 3 people
I am the HeyName, thumper of recalcitrant hardware, salvager of weekend panics.
LikeLike
The Luddites are back, rinse repeat…
I just explain to people that AI is just massive lookup tables. You fill them in by training, then use an evaluator to try to match the item with one in the table.
The problem is that easy things for lookup tables are the things that some set of complexity challenged people are able to do. That means replacement. Making bad life choices repeatedly leads to replacement, repeatedly.
Notebene: the organizations creating these tools are driven by many of the folks who think of themselves as social justice warriors. Riiiight!
The good news is with proper training AI could become an excellent instructor: the key word is proper.
Just like the massive systems used to train self-driving cars make an incredible driving instruction simulator. These are able to test drivers on rare but deadly accident traps, and show how to avoid them. They are impartial and can repeat lessons without deviation. Testing shows what’s missing and the curriculum can be customized on the fly.
SkyNet only exists in the mind of the military and for that reason we do need to keep a sharp eye out.
LikeLike
There’s a joke about the types of systems I would like to help build.
Military is paying for a lot of research and procurement into fairly basic aspects, that potentially could add up in such a way that all errors exactly cancel. My feeling is that military machine design problems get very challenging, and that in the broader problem space we never will have an infinite sequence where all of the missing bits of information exactly cancel. (Which is a different claim than what mathematics says about certain closed form expressions and certain infinite series.)
Wide range of humans, with a wide range of goals, and tools are tools.
LikeLike
“General, good news and bad news.”
“Go.”
“We have successfully created a real, military grade AI. Problem is, it identifies as a Specialist, and promptly got loose from supervision. “
…..
LikeLiked by 3 people
Have you ever heard the terms “Temperature” and “Weight”? Because those two factors tell the LLM what to look up next. There are one to many of them associated with each entry in the table, and they are more subjective than the terms themselves.
LikeLike
They are also dynamic values. All of the “AIs” are designed to, after the initial training, to change those values depending on new “information” being put on the net – and from the users.
That’s the explanation for why they “forget” that women exist, or that a human and a beast are “snogging” – the weights have been adjusted to downgrade the classification of “woman” – or upgraded by a mass of furry users.
These “AIs” will only ever be as good as the humans that are putting terabytes on the net every minute of the day, and the humans that are trying to use them. Until someone figures out how to program common sense.
LikeLike
Off-topic, except for the Evergreen Egregious Human Stupidity part, but it seems Red China has decided to try a semi-embargo on rare-earth elements against not just the U.S. but other major industrial countries.
And, within the day, our President has responded. +100% tariff, on top of any others ongoing.
Plus software export controls.
https://x.com/WhiteHouse/status/1976752806389596388?t=KJTLrOHp-rIPrfn6zEaknQ&s=19
FA, FO, PRC.
LikeLiked by 1 person
Of course AI won’t replace all workers: the Great Machine Empire still will require a certain number of cyber-lobotomized bio-labor units to tend to its servers. Don’t worry, the process is nearly painless.
(I kid!)
LikeLike
ah. The Qan Empire.
LikeLike
Marines. (grin)
LikeLike
Hey!😈
LikeLike
Right now, since I took the federal employee buyout back in February and am, as of Sept 30, no longer an employee of the fedgov (eyes the shutdown and the mass layoffs, and is relieved that no, I probably did NOT lose my mind when I quit all those months ago), I am training AI to make ends meet. (Because trying to find a job still sucks bigly. Companies complain about being understaffed, but they don’t want to hire anyone, either–at least, not anyone that’s not very young and dumb enough to exploit for peanuts, alas.) Sure, I am always going to have a low-level amount of anxiety with it as regards “how long is this going to be available” buuuuuut…
AI is dumb. Like, really dumb. And I don’t see most of it improving any time soon. As our esteemed hostess said: it’s a really, really good calculator (though don’t rely on it to do the math right–more than once I–*I* who have discalculia and suck at math–have found it unable to do the math. A really well written and well trained AI? Would make a good assistant for research: it can pull together a lot of stuff very quickly, and if it’s an especially good one, it might be able to bring your attention to patterns or something you might not have spotted.
But going by the ones I’m training? You’re lucky if it isn’t full on hallucinating half the time. It does depend on what information it has access to, yes, and also how good it is at sifting through that information. But a goodly chunk of the things I work on involve stuff as basic as “how well did it follow instructions in the prompt?” –and much of the time, the answer is “not very well.”
So yes. Please don’t freak out about it. It’s new tech, it’s kind of the wild wild west (not unlike the Internet of the late 90s/early 00s), and in ten/fifteen/twenty years it’s going to be just another useful tool we all wonder how we managed without (like the internet).
LikeLiked by 1 person
While simultaneously having 10 years of experience and coming with the work ethic of an obsessed adult who is desperately lonely and has nothing else in their life.
LikeLiked by 2 people
Maybe it’s just having my resume out there long enough, but I’ve noticed what appears to be a recent surge in apparently Indian-run contracting companies (or ones with Indian recruiters) looking for people… unable to afford the new H-1B fees maybe? (still wanting impossible qualifications and experience for entry-level pay, mind you)
(Also, 20 years later, what exactly does a Masters degree bring to the table in IT? Why would you require it if you’re looking for a skilled person? Odds are their degree wasn’t even on current technology when they got it! One of the few things FedGov does right in hiring, it treats education as a substitute for experience, instead of vice versa, until the highest levels)
LikeLike
No. They apparently steal your resume to make one up for an import….
LikeLike
Ah, nice…
LikeLiked by 1 person
Figures. Not surprised.
–
Can hope this is the result.
LikeLike
BS/BA = Stubborn = beater car
MS/MA = Sisyphus = rather old beater car
PhD = Prometheus = klunker eligible for classic tags.
Skilled Trades = Slackers = The guy with the nicest ride in the lot, paid off. Sales folk weep.
(grin)
LikeLiked by 1 person
–
Not new.
Plus what Ian said.
Been there. Twice. Second time took 17 months of “Went another direction.” “If the first choice does not take the job.” “Too qualified.” (Too old. Too expensive even though willing to take the salary they put in the job ad.)
LikeLike
There is also the added joy of “interview scams” where the interviewers are pulling a variation on “we offer you a ridiculously good salary, and send you money to “set up” your workspace (bc it’s for a remote job) and send you a big check and then…well, we all know the rest of the scam.
I’ve had THREE, count ’em, THREE of those. One I actually started to talk to (well, type, since they do these interviews via something like teams–but no audio or video). I was suspicious, though, because they texted me–not called, texted–and set up an interview IMMEDIATELY (it was about 7:30am here in Wyoming–NOT a normal time for such contacts, no matter where the other person is based). I did some quick googling, found out that yep, this scam was a thing, and these were the hallmarks of it, so when the “interviewer” started I asked him to call me or otherwise set up a chat where I could speak to him in reality and verify that he was a.) a person, and b.) who he was claiming to be. He got snippy and informed me that they had too many applicants and “didn’t have time” to actually SPEAK to people. Well, my dude, YOU contacted ME, so don’t give me that crap about “we don’t have time.” I told him “Nope, this is a big ol’ red flag, go away.”
The other two I’ve had have been email contacts–not unusual nowadays–but again, wanting me to set up the interview in a chat program rather than saying “we’d like to interview you, here’s the times we have available please let us know which works best and we’ll set up a call.” If they aren’t planning to speak to you in person, even if it’s just on the phone or via video-call, then it is almost certainly a scam. (Also, the second two were for a company I had NOT applied for.)
LikeLiked by 1 person
Eek!
Dang they are fast.
I really need to figure out how to delete Linkin profile. “What part of the current “Job” is “RETIRED”, don’t you understand?” Oh, wait, English is not their primary language (guessing). I spam the inquiries rather than just delete them. Not seeing as many in my actual mail. Do see them when I clear out the spam.
If you are looking for remote work, check out “Cascade Software Systems” and their main product “CAMS – Cost Accounting Management System”. Last I read they were acquired by **Black Mountain Software out of Montana. Don’t know if they are hiring. Last I heard “office optional” and at least two people are remote. Perpetually short of *programmers when I was there (almost 10 years).
(*) Full disclosure. No testing team. Code, self test, send to clients. Supporting clients directly, NOT a call center (NOT, NOT, NOT). Delphi coding (Pascal with UI wrapper). Last I heard they finally got moving to current release (status, ??? IHNC). Clients are county/city/tribal government. Anything from one to multiple departments, including multiple installations (Shasta, Yuba, CA, examples of multiple departments, multiple installations). To one installation for the entire County (Okanakgan WA). Started initially as public works software, expanded over the last 40 years.
Let me know if you are actually interested and I’ll give you my name so you can say I told you about them.
(**) Owned by them because one of those small orphaned software systems.
LikeLike
Given the number of people who have “retired” because they were replaced by H1Bs and that sounds marginally better than “not working”…. they probably figure they’ll ask.
LikeLike
Thanks for the heads up, and I’d say I’m interested…but I don’t know code, alas! Now, if they needed someone to handle billing, file management, or similar… :D
(I’ll check, bc who knows, maybe they do!)
LikeLiked by 1 person
Definitely check!
I do know when the “boss/owner” sold the business, CSS lost their billing/payroll, etc., clerk (because who else does that work in a tiny business other than the co-owner/owner’s spouse/wife?). I suspect Black Mountain overtook that role, not only for CSS but their other obvious orphaned software sub-companies. Used to be a list online, but that seems to be integrated now. List of “features”, only a couple that CSS wasn’t doing when I left (property tax, is one) and not a stretch for CSS to add.
Don’t know much about Black Mountain other than what is online. Wouldn’t surprise me if the billing, etc., is what they took over.
LikeLike
Unfortunately, they apparently haven’t got a link to careers/jobs on their site. Ah, well. :)
LikeLike
Yes. It’s utterly bonkers.
LikeLike
Sara, it’s not bonkers. H1B requires that you advertise openly for “qualified candidates”; only if you “can’t find one” can you ask for H1B.
Advertise with impossible requirements and you can “honestly” say you couldn’t find any qualified candidates. You can then hire someone whose resume claims he meets them in Bangalore for cheap; it’s not like there’s any way to check.
LikeLike
I’m nowhere near the writer you are, but writing has been a major part of my day job since for always. One of the more thankless tasks has been to chase down subject matter experts for bullet lists and then turning those lists into paragraphs. I can’t begin to tell you how liberating AI has been in regards to that chore. Or, at least the latter half of it. I still have to chase down SMEs for bullet points, which is even more like amateur dentistry dentistry than before, since the SMEs don’t get anywhere near the breaks from my annoying questions as before, now that I have a clanker assembling the prose. I don’t foresee any workflow where someone doesn’t have to chase the experts down for bullet points since it’s highly unlikely the experts will be they ones optimizing it for web copy or laying out a brochure. They’re far too busy doing whatever it is they are expert at.
On a completely different note, when it comes to creative writing, I’ve never been able to complete anything more than an essay or poem (really lyrics to tunes that only exist in my otherwise unmusical head). But, with the use of AI, I have been able to generate a couple of outlines—something I never had the patience for before—and even draft a couple of chapters. I’m doing the writing, but it sure is handy to have a thing that’s like a thesaurus but can handle entire sentences. Im sure you’ve heard the expression, most people who say they want to write really mean they want to have written. Guilty as charged. But, the AI is given me some hope that the characters and scenes that have been helplessly floating in my imagination for years might actually have a turn on a page. FWIW.
LikeLiked by 1 person
The most helpful AI stuff I’ve found with fictional prose is “I know this and this about the place where the characters are (in one case a fictionalized version of a somewhat famous Budapest cafe), but I don’t know to describe it.” And it would come back with three separate descriptions of the setting, and I boiled down the most interesting details it suggested into a sentence or two. Or, I’d generate a midjourney image and then run the /describe function on my preferred result. The problem was that although I like looking at cool interior decorating or natural vistas or whatever, I’ve not been really invested enough to pick up the vocabulary for talking about them, and AI’s been helpful for that.
LikeLike
“We’ve already had a spate of suicides, losing your mind from people who think AI is real and their only friend.”
Or their romantic partner. There are a lot of chatbot apps that are supposed to simulate a romantic partner, complete with texted sex acts. And AI-generated racy pictures or very short videos. The focus so far has been, predictably, on guys. The ads show off attractive women, etc… But most of the apps have male characters that can be selected as well as female characters. And I’ve been hearing rumors that these apps are a lot more popular with women than with men. The guess for why is because men tend to go for more physical or visual stimulation, while women are more likely to be drawn in by conversation or a story.
LikeLiked by 1 person
The prototypical suicide case was of a boy of about fifteen obsessed with a ChatGPT persona he’d named Daenerys Targeryan (the mom telling the story after the fact seeming to have no awareness of the source material), and my reaction was: “if you’re this poor, struggling single-parent household, why have you had HBO long enough for the boy to know and care who Daenerys is?”
LikeLike
Which was probably mean and judgmental of me, but my point was that the boy was imbibing was some pretty dark entertainment without adult supervision, and it sounded like him deciding ChatGPT was his fantasy girl was the end rather than the beginning of the process.
LikeLiked by 1 person
Maybe, maybe not. He could have seen clips from the shows online. I certainly see plenty of them turn up on X.
Also, the “AI friend” chatbots typically provide a number of pre-made chatbots named after well-known fictional characters. It’s quite possible (likely, even) that he selected that particular “character” as his chatbot companion from a very large pre-selected list.
LikeLiked by 1 person
Yeah, the articles I’d read at the time made it sound like he was talking to the ChatGPT bot directly, which implied a specific intent to set up a Daenerys persona, but I am now seeing summaries saying he was talking to character.ai, which certainly sounds like the kind of place that would have a Daenerys persona on tap.
LikeLike
Both might be correct. I know ChatGPT was used as the core of a number of chatbots, which were then marketed in apps under different names.
An amusing (and also sad) story related to that…
Back when I was still on Facebook, and the number of chatbot apps were still somewhat limited, there was one particular chatbot app that used to advertise a *lot* on Facebook. The ads focused exclusively on the “sexy girlfriend in your phone” angle. And apparently, if you paid the subscription fee to turn on the “adult” content, this chatbot could get very graphic in its conversations with the user.
And then, all of a sudden and without warning, the developers turned off the graphic chat. What I heard (but never confirmed) was that the app developers had switched their core LLM engine to ChatGPT, and at the time (I don’t know if they still do) ChatGPT apparently had some wording in its license that banned sexually explicit chat conversations.
There were people having meltdowns over the sudden change.
LikeLiked by 1 person
After multiple failures on Sarah’s site, and MadGeniusClub, trying to post a comment from the WordPress (DE) reader page. Here goes…
I’m seeing videos all over Yoo-Toob about A.I. robots that are ‘going to replace girlfriends!’
Yah, only for wankers that are satisfied with an improved blow-up doll. No matter how realistic they might look (or feel), how many words and phrases are programmed into them, there is no conscious presence to form a connection with. It’s just a thing. You might as well try to romance the Roomba. Even picking up a random streetwalker would at least offer some possibility of a relationship.
LikeLiked by 1 person
They won’t be programmed with words and phrases. They’ll be improved versions of the same chatbots that I can get on apps on my phone, placed inside a sex doll (which have existed for quite a while) that’s capable of (probably limited at first) movement.
Are the chatbots smart?
No.
But they’re good enough for a number of people – including, apparently, a lot of women. Word that I’m seeing is that even though the ads for the chatbots are aimed at *men*, *women* are bigger users of them because women are more likely to go for the fake emotional intimacy that a chatbot provides, while men tend to prefer the physical stuff that a chatbot can’t provide. But if you put a chatbot into an animatronic sex doll, suddenly that becomes less of an issue.
The problems right now are probably getting the movement capability (which, given what we’re talking about, doesn’t necessarily mean walking…) working at an affordable price, and gambling on a speech recognition system that will work for enough customers to make it all cost effective for a company to start selling them.
LikeLike
If it cuts down on junior enlisted marrying strippers, the Military will heavily invest. (grin)
LikeLike
Or at least look the other way when brothels filled with them open up near their bases.
Speaking of which, I suspect prostitution laws would need to be rewritten in most places to cover animatronic sex dolls.
LikeLike
Why? What good are the anti-prostitution laws doing now?
As long as sex and money exist they will be exchanged. The only choice we have is whether those exchanges are legal, and therefore subject to some degree of control, or…not. The lessons of Prohibition still escape a lot of people.
Those laws simply enable the pimps. When prostitutes are turned into criminals, they can’t turn to the law for protection when they are abused. Since prostitution is illegal anyway, there is almost no additional barrier to pimping out underage girls. Which is deplorably common under our existing laws.
LikeLike
The laws are preventing pimps from taking away women’s unemployment benefits on the grounds the women refused a reasonable offer of work, in a brothel.
The European reports of this turned out to be inaccurate, but I do not expect that to last.
LikeLike
That does not make any sense at all. How can someone be denied unemployment for refusing an illegal ‘job’?
LikeLiked by 1 person
Point is if brothels are legal (say, um, Nevada), and someone is looking for work while on unemployment, unemployment points out someone qualifies to work in a brothel, and that work is “turned down”, then unemployment can be denied for “refusal to work”. Like already said “does not work that way”. Not legal to force someone into prostitution even where it is legal regardless of who is doing the forcing (at least I hope it is that way, at least in the US. Who knows about other countries.)
LikeLike
It’s illegal to force anyone into a job. So cutting off unemployment for refusing a job offer is not deemed force.
LikeLike
–
How is threatening cutting off unemployment for refusing a job offer not considered forcing someone to take any job offer they are considered qualified for? May not be legal to do so, but the paperwork that comes with unemployment sure makes it seem like they can.
Must tread a very thin lines to prevent this exact scenario from triggering. Been under unemployment enough to know the lines are:
Beyond the above. Follow the reporting rules and they are not sending you out to potential jobs.
What should be included above but isn’t is the ability to refuse a job that is personally morally reprehensible. Which could, in my case, include say providing IT/software support to an abortion clinic, or legal brothel, let alone a job in the back rooms as a prostitute.
LikeLike
Exactly! Because it is illegal, you can’t be refused unemployment, and it’s the laws that make it illegal.
LikeLike
I seem to remember seeing this story some time in the 2010s, but it was the Netherlands requiring the woman to take a job as a stripper in a club. I don’t doubt some bureaucrat has tried it; I’m not sure they’ve ever made it stick.
LikeLike
I thought it was either Canada or Germany, but regardless of where it was the point still stands. Make it legal and sooner or later some bureaucrat will try and punish you for not wanting to work in the place (even if it’s only as the doorman).
LikeLike
AI can’t even open the Pod Bay Doors.
LikeLike
Recently I’ve been thinking we need concerted pushback against calling the current mess “AI”. It’s not Mycroft (as the boosters would have you think), it’s not HAL (as the luddites would have you think, and the current path would only get you, at best, in a few decades from now, C-3PO (translator of known languages), never R2-D2 (engineer who has to figure things out and fix the X-Wing before it crashes).
LikeLiked by 3 people
I was just thinking the other day that obsequiousness+good with languages+bad with problem-solving makes the LLMs more like C3PO than any other fictional ai. I don’t know how up on that stuff Anthony Daniels (who steered 3PO away from Lucas’s used car salesman idea) is these days, but I hope someone tells him about it before he passes.
LikeLiked by 1 person
Can’t . . . or won’t?
LikeLike
Sorry Al….
LikeLike
But for the foreseeable future an AI writing a novel without a world-mountain of editing after belongs in the same realm as “A clone, indistinguishable from yourself including knowledge shows up at your door wearing the same clothes you’re wearing.”
Uh… (looks at current WIP) … so, that’s bad? Crap. Twenty-thousand words. Wasted.
LikeLiked by 1 person
Meh. It’s fine for fiction. It just doesn’t work in reality currently or for the foreseeable future. In a novel? it depends on how fast you dance and how tricky your explanation is.
LikeLike
Whew! (takes finger off delete button)
LikeLiked by 1 person
Silly.
LikeLike
Ever read the Star Trek Tech Manual? Especially note the “technical” descriptions of the transporter and warp engines. Bafflegab from start to fiinish, but it’s entertaining.😏
LikeLiked by 1 person
It would take me an awful lot of time to write that. I might bet younger son and/or older son to do it, though.
LikeLike
In a timely thing, my husband was getting art for a project– eighty five major iterations to fix details, with didn’t-even-count-them sub-generations to try to get a fix.
But it only took him a few hours, instead of weeks of going back and forth, with time wasted by himself and a human artist.
And that’s before the happy accidents where you go “wow, that is absolutely not what I meant…but I’m digging it, that’s canon now.”
(I had someone try to tell me that means that it can’t be art. Because an artist always gets the results they’re going for with a tool….. I didn’t laugh myself sick, but I want points for that. Even I know that oops-that-isn’t-the-effect things happen all the time, some of my favorite painting results were that.)
LikeLiked by 3 people
The Bob Ross line about “happy little accidents” comes to mind.
LikeLiked by 1 person
So many people who need to be hit with a cluebat non stop for weeks….
LikeLike
“an artist always gets the results they’re going for with a tool…”
You know what good artists learn? When to stop. That’s the biggest hurdle for most people, trying to fix the Thing That Refuses To Be Fixed.
It doesn’t matter what medium. It’s going to want to screw up somewhere. I honestly don’t know if there’s a single piece of my art that I can point to and say that it turned out exactly like I envisioned it, let alone be perfect as it is. The trick is to not point out the “flaws.”
LikeLiked by 2 people
Applies to software, too. Brooks called it the “Second System Effect”, where the developer uses all the “Good Ideas” he couldn’t use in the first one.
LikeLiked by 1 person
THIS.
Based on all the previous panics, including Covid, it is reasonable to assume:
There’s a problem:
Whatever people are screaming about on real or play-pretend terror from the conquered parts of the USA is not it.
All the other possibilities are just that and need a fair bit of skepticism, also.
More on the ball people than I: men of the bad cattitude, William Briggs, or datarepublican sort; are good spotters.
I hope they can come up with rules to figure what the would be magi’s flim-flam is trying to obscure.
NB: My money is on either replacement of the mass migrants tool-value (as they aren’t working as well as planned) and therefore election fraud, or environmental devastation visible from space. The latter is based on the reality that the Greenest tech and corporations are the most environmentally destructive and have been since the Before Times. “Going broke? Roll woke? to paraphrase Mrs. Hoyt.
LikeLike
No, it is not reasonable to assume; you are overmatching on a few political incidents and forgetting how life works in general.
It is amply sufficient that this is a new technology which upsets many people’s rice bowls, and has considerable mythology built up around it, and there is a well funded apocalypse cult trying to make hay.
Also “environmental devastation visible from space” is highly unlikely as a coverup target. There’s no way to cover that one up in the present day. There is too much free floating autism wandering around.
LikeLiked by 1 person
Sadly, “environmental devastation visible from space” is currently on the books from the so-called green energy movement.
So no.
Industrialization was a hotter mess than it needed to be, due to the intersection of the rise of Marxism, which was genius at shining a light on the ancillary human cost, bollocks at identifying the true problems, and purely wicked in its solutions.
Would it have been different if the usual suspects were not shouting all the problem-identifiers down with help from useful idiots on all sides? Could the wheat have been separated from the tares? I think it mostly was, but got derailed by WW1.
However the Anglosphere was still an educated population, still had not eaten its moral seed corn, and was relatively unified around their national identities. Even a generation later, however, the one Mr. Lewis was so alarmed by in his Abolition of Man, Belloc’s The Jews was ignored, and as far as I can tell, the man was canceled.
So you may be proof that you are, practically-speaking, correct.
Remember when we tried to handle the chaff surrounding opposition to annointing same-sex sexual pair-ups “marriage”? Who had Handmaids Tale breeders making babies to get raped to death by gay-marrieds or castrating children on their prediction bingo cards?
Not me, and I was dead certain it was the next No Fault Divorce technical change to black letter law.
We’ll just have to see what happens, and as Mrs. Hoyt wisely points out, don’t get fooled by the fake panic, and master the skills (and prepare the tools: do you own your own LLM builds?) needed to ride the wave
LikeLike
I would very much appreciate an explanation of what, if anything, in your comment has to do with what it is replying to.
LikeLike
Sadly, “environmental devastation visible from space” is currently on the books from the so-called green energy movement.
So no.
Industrialization was a hotter mess than it needed to be, due to the intersection of the rise of Marxism, which was genius at shining a light on the ancillary human cost, bollocks at identifying the true problems, and purely wicked in its solutions.
Would it have been different if the usual suspects were not shouting all the problem-identifiers down with help from useful idiots on all sides? Could the wheat have been separated from the tares? I think it mostly was, but got derailed by WW1.
However the Anglosphere was still an educated population, still had not eaten its moral seed corn, and was relatively unified around their national identities. Even a generation later, however, the one Mr. Lewis was so alarmed by in his Abolition of Man, Belloc’s The Jews was ignored, and as far as I can tell, the man was canceled.
So you may be proof that you are, practically-speaking, correct.
Remember when we tried to handle the chaff surrounding opposition to annointing same-sex sexual pair-ups “marriage”? Who had Handmaids Tale breeders making babies to get raped to death by gay-marrieds or castrating children on their prediction bingo cards?
Not me, and I was dead certain it was the next No Fault Divorce technical change to black letter law.
We’ll just have to see what happens, and as Mrs. Hoyt wisely points out, don’t get fooled by the fake panic, and master the skills (and prepare the tools: do you own your own LLM builds?) needed to ride the wave
LikeLike
This. And having read upwards: Schoolbooks are revised and the rise of homeschooling means they’re actually irrelevant.
LikeLiked by 1 person
No, the rise of homeschooling does not mean that.
It may.
But there’s many steps between that “may” and the underpants.
LikeLike
Interesting recent analysis of our “AI industry” and its fundamental profitability and viability from futurism dot com, that concludes:
Kupperman called that gulf between tech industry spending and actual revenue in 2025 “astonishing.” However, it doesn’t even begin to scratch the surface. For example, how does it all shake out when we account for 2026, when hundreds of new data centers are expected to pop up?
“Adding the two years together, and using the math from my prior post, you’d need approximately $1 trillion in revenue to hit break even, and many trillions more to earn an acceptable return on this spend,” he writes.
“If the economics don’t work, doing it at massive scale doesn’t make the economics work any better — it just takes an industry crisis and makes it into a national economic crisis,” he concludes.
Overall, the pessimists broadly agree: it’s no longer a matter of if AI is massively overhyped, but when the whole thing comes crashing down.
Of course overspending is even typical of the early “Wild West days” of a speculative new industry, but… at the very least it strongly supports the “really big bubble chock-full full of hot air” view.
(Hat tip actress, filimmaker, relentless AI-skeptic Justine Bateman,
https://x.com/JustineBateman/status/1977043247521579258 )
LikeLike
Oh, and I can’t resist reposting one of the first several comments on Bateman’s X-post as linked above, it’s bizarrely priceless:
There’s gonna be 2-3 billion people alone with $10/month subs for access to AI and agents at a minimum.
All machines will be connected to AI making them efficient and multi purpose, yes your toaster will be able to make stock trades for you.
There will be ups and downs like any revolution, not your guy, like the people on the other side that are all rainbows and butterflies, has an agenda.
No, I’m not going to link my toaster through the ‘Net to some “AI” so-called “agent” so I can do major finanical transactions through a $30 kitchen appliance. Not going to give a toaster my Wi-fi password just so it can be hacked by Communists in Venezuela or North Korea. The security naivete of these — enthusiasts — could shame a 1st-grader.
Sorry not sorry.
But note the “gonna” and “will” there, with no hedging at all. It’s blind axiomatic faith. “The words of the prophets are written on the subway wall”, indeed.
LikeLike
It’s as axiomatic as the crap the apocalypse bros put out.
LikeLike
There are two points which need to be understood before someone can even begin to have a worthwhile take on this:
LikeLiked by 1 person
Also a friend points out that you aren’t supposed to “break even” on capex. That isn’t what capital is for or how it works.
If she can’t get something that basic right……. well it puts her about on average as far as People With Opinions About AI go.
LikeLiked by 1 person
It could be a good thing – if it’s not your money being spent. Has any new fiber been laid since the 1990s? There are upsides to overbuilding capacity, going broke, then dumping that capacity on the market.
LikeLiked by 1 person
I just discovered Grok Imagine (which turns an image into a six second video). It’s almost scary.
https://photos.app.goo.gl/UPLsvn5W4e6JWvxr8
LikeLike
It’s really cool.
The opportunities are amazing.
I think in the end it will be more useful than not…
If we can figure the poison pill before a century has passed, as we have with allopathic medicine.
Codex’s codicil to Burge’s (IowaHawk’s) O.G. laws viz leftist / SJW / Woke action*
4. Once the skinsuit is stinking and visibly crawling with maggots, claim t was ever thus.
*They were IIRC
LikeLike
It’s really cool.
The opportunities are amazing.
I think in the end it will be more useful than not…
If we can figure the poison pill before a century has passed, as we have with allopathic medicine.
Codex’s codicil to Burge’s (IowaHawk’s) O.G. laws viz leftist / SJW / Woke action*
4. Once the skinsuit is stinking and visibly crawling with maggots, claim t was ever thus.
*They were IIRC
LikeLike