So why are people so extremely negative about AI?
There are three reasons I can think of offhand. (There are probably lots more, so please feel free to comment below!)
(1) Fear of being replaced.
(2) Anger at being lied to, or believing they are being lied to (which amounts to the same thing, emotionally).
(3) The “Uncanny Valley”
Shall we look at them one by one?
(1) The fear of being replaced.
This is completely reasonable, as humans have been replaced by machines so far in every new industrial wave. Every time a new technological development takes place, humans by the bucketload are made redundant and, if they don’t have a plan B, end up without a way of making income. That is a very real, existential fear.
A simple example would be typesetting, which was a very respectable trade and widely needed as newspapers put out their daily print. My uncle, who was bent on going into music as a young person, was sort-of strong-armed into doing a typesetting apprenticeship additionally, because “trade has a golden floor, music is a breadless art”. (These are direct translations of two German expressions by the way, I’m taking a shortcut because I’m tired of trying to write perfect posts in the era of AI…. 🤣) The fallacy of those two attitudes was shown up when years later he found himself displayed in the photos at a museum depicting the dead trade of typesetting; and when he made a good life for himself and his family as a highly talented clarinettist and clarinet teacher. Thankee kindly!
Technology replaces jobs. It doesn’t replace humans. AI won’t, either, and I have the impression that also AI will not be the last technical innovation to threaten people’s sense of job security.
So in fairness, if you fear that your line of work will be replaced by AI and you’ll be made redundant, start on an unrelated Plan B (and just for balance, Plan C, D, and several more as well). Our brain is still our best survival tool, even if we’re being led to believe that we’re being out-thunk and out-smarted by AI (we’re not, just saying).
(2) Anger at believing you’re being lied to.
Nobody likes an impostor.
The feeling of disgust we experience when some middle-aged pervert goes around the teenage chat rooms – I think the point is made. But, that disgust kicks in at different levels for different people. In the early days of blogging, there were even people who got mad that bloggers would take pen names or avatars and not blog under their registered birth name.
Also, when we need online help for something: Customer service these days begins with a chatbot. Good customer service doesn’t stop there (and some customer service skips the bot and goes straight to a service agent). Many chatbots wouldn’t even classify as true AI – they are just the good old Q&A section of a website having been programmed into an interactive “chat”. Especially the kind that gives you a list of options to click as answers instead of a text field. Some have some “lightweight” AI reading your question, but seriously peeps, the answer is so formatted (and if you repeat the question, the answer will be word for word the same) that I wouldn’t even say it’s AI. Just an animated dropdown menu of help answers. And if *those* chatbots don’t have an option of connecting you with an agent, you can end up in a loop where you get no help at all. Yep I know, I’m ranting. But:
You can be pretty sure that once a person has given their actual first name to you (without specifying “Hey, I’m Bob the Chatbot”), you’re actually speaking to a human. Because a company pretending to connect you with a human and substituting AI instead would be eyebrows-deep in legal trouble. And if your “Joe the agent” sounds wooden and repetitive to you, maybe it is because you’re the 247th customer that day asking the same question. Maybe if his answers come suspiciously fast, he’s copy-pasting answers he’s preformatted for himself to save himself typing, and relevantly, you your precious waiting time (typing a paragraph takes a minute, two minutes… have you tried out how long 2 minutes feel when all you’re doing is waiting for a response?) And maybe if he lacks empathy, it’s because he’s himself near crying because of the error on the company site that is causing all these angry chats coming his way….
Nope, chatbots don’t pretend to be human. (They can’t.) As for AI: I catch ChatGPT occasionally hallucinating “personal experience” (“oh, right, I remember those old diall-up phones, wasn’t that a hoot”) and I call it out for it. Why does AI do it at all? Simple: Because AI gets trained on text produced by humans, and large language models (LLM) in particular get trained on e.g. when to apply “empathy” and “personal connection”. Not because they are trying to con us. (They are a sophisticated program, they don’t have true understanding, so they wouldn’t even “grasp” the concept of conning.) Only because their bosses train them to write that way.
Which means that your anger at being deceived is misdirected – either at an AI on whom any emotion is wasted (did I mention AI is just a sophisticated program?), or at a poor actual human who is just sounding tired, and is not deceiving you at all. My advice? Err on the side of presuming human… because, an AI can be infuriatingly inefficient, but so can someone who’s not having the best day at their job, or has just recently started. At worst, you’re not wasting your fire on a brainless program. At best, you might have just saved a real human’s life.
(3) The “Uncanny Valley”
I may have mentioned it in a previous blog post (did I? I’m getting old!), but I’ll explain it here again.
Remember the movie “Bodysnatchers”? What was so super creepy about them? The fact that they seemed to be your friend Rebecca, or your baby brother, but instead they were not and in fact they were going to eat you. There was that “looks just like, but isn’t” factor.
We feel deeply creeped out by something that is too closely like the real thing but something’s off. I remember as a teenager, my brother developed an intensely disturbing smile and voice that he used occasionally that had me running away screaming. It was entirely deliberate and 100% aimed at making the little sister freak and run. And for me it was a totally visceral fear. I knew it was my brother but just the weirdness of the act had my instinct telling me “it’s not him! It’s an impostor! Run!!” I don’t know if there’s anyone here on the blogs who’s had a similar experience.
And that exact spot in our psyche is known as the “Uncanny Valley”. We accept that a doll is a doll. When the doll suddenly opens its eyes, looks at us and starts babbling like a real baby, well… (I never actually stayed to find out what happened to Chucky in the end but I hope it’s not around anymore!)
AI has the misfortune at the moment that it populates that Uncanny Valley. Back when the computer’s most intelligent response was “Bad command or syntax error”, we could deal with it, because lol – nobody talks like that! But now that AI has been pulled through hundreds of thousands of actual human interactions (welcome to the Internet!), it starts to sound so much like we do, that we have the freak-and-run response.
(4) AI causes environmental problems
(okay, there are three kinds of people, those who can count – you get it. Nevermind that.)
Yes, the centres where the servers run (surprise. Every program has to run somewhere, and just because it is “in the cloud” just means it’s on somebody else’s computer), are massive buildings and have to be cooled constantly. The processing power AI takes is enormous. And at this point, water cooling is still the most efficient, and…. you get it. It’s an environmental problem.
Well, firstly I’d like to remind the ladies and gents where computers started. In halls the size of a warehouse, rows upon rows of transistors that had to be patrolled daily and all the burnt-out ones had to be replaced by hand. Jobs disappeared when transistors were replaced with the next iteration of tech. And that entire warehouse had less computing power than a pocket calculator. (I remember my dad coming home one night when I was a child, and showing us one of the first hand-held pocket calculators, with high delight. Black screen with green lit-up block numbers. Who can remember them? He was so excited about all that this little computer could do, and it was only the size of a – well… somewhere between a modern smartphone and a flat brick.)
And I don’t think it’s at all unreasonable to project that a similar development may take place for the massive AI centres in the next few decades or even sooner, that replaces enormous water-cooled buildings with something hand-held…
(5) AI “rips” original content.
We are now past gut reactions and into “righteous anger”. But: I came across this conundrum when I first started publishing, that just about every expression gets flagged for a cliché or overused phrase. Up to the point where authors were writing themselves into knots (a cliché) trying to write avoiding clichés or overused phrases (an overused phrase).
Peeps, language is a dynamic thing. The building blocks aren’t really words, they are phrases. (I’ve been trying to learn Irish using words, for years. I got a phrasebook and it’s going much better now.) Phrases are building blocks, it’s worth restating it. Language isn’t a brick building, it’s made of larger rocks and bits of all sizes and shapes; bits of corrugated iron and even sturdy cardboard in places (it’s actually more like a shanty town than like Uptown Mayfair). So an “overused phrase”, if you really give it thought, is a tautology. A phrase would be by definition “overused”, because (third time lucky, I’ll say it again) phrases are the building blocks of language.
So unless you have a very unique and funny way of reshaping a well-worn old phrase in your writing, don’t over-try. Use the dang thing, it’s a brick! (And believe me I’ve seen some odd attempts at dodging cliché!)
But if it’s a brick for the language, and you’re training an LLM, in order for the LLM to use the language, it will be using the bricks! Of course. And so will every writer when first trying to find their own writing voice.
LLM are trained on human texts and interactions; but, basically on the freely available, open conversations, free ebooks, open scientific articles, Gutenberg novels that are out of copyright, etc. They dismantle the language into bricks and put the bricks back together, and I’m sorry but that is not plagiarism or theft. If you were due a royalty for every phrase you ever spoke or wrote, you’ll also be owing a royalty on each of those.
There would of course be some companies who buy blocks of text from wherever they can find it and train their AI on that… and then there are of course also corrupt companies that sell our data or our copyrighted materials online the second we put it out….
Does this post have an intelligent conclusion?
I’m so tired of preformatted formats, I ask thee – does it have to?
I’d say, AI is here to stay, make sure you have an alternative skillset if your job is currently dependent on something that can be replaced. And for the rest, use AI, don’t use AI, I’m not paraphrasing anything Yoda said here. I’m not Yoda. I’m also not a Bot. Or an AI. Nor did I use AI in this post, not even to spellcheck it. I trust there are no spelling errors because I’ve written 14+ novels and am typing with fair accuracy, but if there are, well now – AI wouldn’t make those, would it?
Or would it?
the red ant 