11 things I hate about AI

Behold! I have had a thought.

Hi!

Hello!

There are two wolves inside of me. One of them nods politely while reading about progressives' refusal to engage with AI being counterproductive, and the other wails like a toddler and screams "I hate it I hate it I hate AI I hate it make it go away". Intellectually I agree with the first wolf, but every fibre of my being wants me to run off with the second one. Why is that?

It's the question I asked myself last week, and I decided to start making a list. I understand that simply going "I hate it I hate it I hate AI I hate it make it go away" is off-putting, especially as hundreds of millions of people across the world use ChatGPT as it is. That's why I felt that trying to put my instinctive fear and disgust into words could potentially be helpful.

If nothing else, it may help me formulate my arguments better. At present, when I encounter people who say things like "I think generative AI has the power to change the world for the better", all I want to do is go "wahhhhhhh" in response. No-one has ever been convinced by "wahhhhhhh", much to my chagrin. Waluigi would be the franchise hero otherwise.

Here, then, is my attempt to put my primal screaming into shape. Why do I hate AI? Oh, let me tell you.*

*I am, to be specific here, mostly talking about regular people using generative AI for…stuff. I think it's obviously fine and good that, say, scientists are able to use LLMs for science now. I've no truck with AI being used in very specific, narrow ways by people who know what they're doing. That stuff doesn't bother me at all.

1. The boomer angle

Look, I'll start with both the easiest and the most annoying point. I think it's good to have to try, actually.

As discussed ad nauseam in this very newsletter, I've always prided myself on how little effort I put into my university essays. Few people tried less hard than me. I don't think I even reread them after writing them. I wasn't sober when I wrote several of them.

That doesn't mean they didn't teach me anything, however. Sure, I'm not sure I actually, truly learnt about the topics I was meant to learn about - please, god, never ask me what I remember from my degree - but I learnt other stuff instead. I learnt how to form arguments even when my grasp on the topic wasn't the strongest. I learnt how to sound authoritative when, honestly, it was a stretch.

I learnt how to structure an essay so it would sound at least vaguely persuasive. I learnt how to work under pressure, even if that sometimes meant getting to the university library at 9pm for an essay due at midnight. Imagine how much more I would have learnt if I'd actually decided to study! My point is, though: the mere act of faffing through my degree taught me a lot.

I just don't think a university student who feeds everything through ChatGPT will gain access to all that. That's a problem because - this is where I start sounding like your dad - there are many areas of grown-up life where you can't really take shortcuts. If you didn't learn how to actually try back when the stakes were reasonably low, and your brain was at its most malleable, you'll almost certainly live to regret it.

I guess the other point in this is that most skills are in some way transferrable. It's easy to think "oh, I have to do X but I'll never have to do X again so I might as well ask the computer to do it for me this one time", but you'd probably be wrong. You never know when X may come up again. You just can't say for certain that it won't. A shortcut may be useful this one time but what will happen next time? You're essentially making a bet that whatever you're doing will never come up in any way again, and I mean it's your funeral, but I just don't love those odds, most of the time.

2. Building something from nothing

So this is something I've spent a lot of time thinking about recently, both in the context of AI and in general. I've been freelance for nearly eight years now, meaning that I've been working without a team or an editor for nearly a decade. I've learnt a lot from this, and may write about it more broadly at some point, but the specific point I keep coming back to is: fucking hell, it's hard to be creative when you're self-propelled. It's hard to come up with ideas when you've got no-one else to chat about those ideas with.

It's a skill I've definitely managed to pick up over the years, but it's still tough. Sometimes, I'll go for days without having a single idea. If you give me a framework, though? Pah, you just watch me. I did the News Quiz again recently and they send you a rough draft of the script beforehand, so you can prepare some little jokes in advance, and man alive that felt like smoothly cutting wrapping paper.

As it turns out, "finding a funny answer to a question" is miles easier than "finding something funny to write about, in a vacuum". Crucially, however, I do think I've only managed to become very good at the former by having been forced to spend so much time doing the latter.

Some proponents of generative AI will argue that going back and forth with the computer helps creativity, as opposed to stifling it: as someone who has to be creative a lot of the time, I just don't think it's entirely true. Sometimes you do need to force yourself to do the hard thing a lot of the time, so that you're then able to do the easy thing quicker.

It's also a good habit to pick up, as starting from nothing is, I think, good for the brain. I see it a lot in oil painting, which I started doing last year. There's nothing as horrifying as a blank canvas, but the first steps you take when faced with it are often the most important ones. Which colour should I use to make my wash? What am I mixing to begin with, for that first layer? Am I drawing some outlines in charcoal first, or just vibing it?

By taking these first few decisions, I start building the foundations of whatever project I'm embarking on. It's a different exercise from, say, paint-by-numbers, and there's a reason why one of those is a more respected field than the other.

3. Learning by doing

Essentially: there are a lot of things in life you can only get good at by doing them, again and again. If you stop doing them, you will become bad at them again.

I personally realised this with mental maths: it's something I used to be really, really good at, until I started spending my days around screens which could do my mental maths for me. I started using the google bar for quick divisions or multiplications, because I couldn't be bothered, and over time I just got worse at them. Perhaps most importantly, it made me lose my confidence.

I used to always back myself, doing sums in my head, but will now, more often than not, end up double-checking the result on google after calculating something quickly. I'm always right, but that doesn't stop me. That's bad! I'm trying to train myself out of it now, but it's taking an amount of effort I wouldn't have needed if I'd not outsourced my thinking to the computer in the first place.

You may, at this stage, be thinking something along the lines of "oh but why would you need to do sums in your head if you've got a thing in your pocket at all times that will do it for you" and man, I don't know what to tell you. I think it's obviously good that I'm able to do these things using my brain. I can tell it's good for me. It makes me sharper. Also: it's only one example, but it's applicable to other areas of life.

I'm just not convinced that "well if there's a way I can not do a thing then I will never do that thing again" is the correct way to live your life. If you really, truly are outsourcing all of this purely so you can focus on Very Important And Fruitful Things Instead then sure, be my guest, but are you? Are you really? Do you think that's what even a small minority of genAI users are doing? I'd love for it to be true, but it's just not what's happening, is it?

4. Think of the billionaires!

Right, harsh change of topic here - segues are for cowards. There's this tweet I still think about a lot which argued that, essentially, billionaires are insane because having that much power and freedom is the cognitive equivalent of getting hit in the head by a horse every single day. We're just not built, as a species, for a life this smooth.

Obviously, generative AI doesn't actually offer its users wealth, but what it does is act as some sort of digital courtier, ever so aware of the fact that displeasing you is the last thing it ought to be doing. ChatGPT will be friendly and it will agree with you and reinforce your priors. It may even say that, yes, the point you made was very astute, and it will generally treat you as someone who ought to be placated at all times.

Do we think this would be good for most people? I don't think it would. I think it's why so many rich and powerful people end up making increasingly poor decisions. It's useful when people hold us to account, and call us out on our bullshit. If your main companion in life is a chatbot which assumes that you must always be in the right, you'll probably end up drowning in your own confirmation bias. In the worst case scenario, you'll also become increasingly resistant to receiving honest but good advice you maybe wish you didn't have to hear.

On a smaller scale, it's something we keep seeing on social media, usually coming from tedious Americans with poorly qualified therapists. Proper therapy, when done seriously, opens up your ribcage with pliers and forces you to reckon with exactly who you are. I've always found it deeply useful, but it's horrible. Proper therapy is like exercise: it makes you feel awful in the short term, in order to make you feel better in the long term.

If you were to listen to those Twitter yanks, though, therapy is this thing you do where you pay someone a hundred dollars an hour and in exchange they tell you that you've never done anything wrong in your life, and everyone around you should probably apologise for all the many awful ways in which they've treated you. You didn't deserve any of it because uwu, awen't you special? Awen't you bwoken in such a unique way that is both fascinating and unbeawably sad? You go and cancel those plans, and feel like you've never owed anything to anyone. I'll see you next week.

While it's easy - and fun! - to make fun of those very online middle class millennials, it's entirely possible that widespread use of ChatGPT will make this sort of thinking more prevalent. Genuine human relationships must sometimes involve tough love: it's the only way people involved can truly grow and become better versions of themselves. Sadly, it's just not in the interest of those AI companies to make their bots less pliable. On which note…

5. Genuine human relationships

They're good! Sorry to be controversial! It's nice to have real, live humans in your life! It takes effort, though, especially once out of school and university. I've written about the labour of adult friendships several times before, but it's worth reiterating: having a proper social circle over the age of, roughly, 30, takes work and occasionally involves doing things you don't really want to be doing, but they're entirely worth it.

There already are a lot of adults out there who are unwilling to come out on a rainy Tuesday or actually make plans and not flake on them, and I'm just not convinced that all of them are happy about it. Chatting to people on the computer and scrolling on social media for hours on end just isn't the same as seeing pals for drinks or dinner. It's nowhere near as rewarding.

The tech barons want you to believe that it is, though. It's in their interest to sell AI chatbots as actual, fulfilling social companions. That's how they'll make their money. We just don't have to trust them, is my argument! We can instead worry about the blatantly vicious cycle that may develop if someone ends up spending too much time chatting away with ChatGPT.

The more you confess to the robot, the harder you'll find it to actually subject yourself to the rawness and unpredictability of human relationships. The fewer friends you have out there in meatspace, the more you'll end up relying on the robot for your socialisation - and so on, and so forth. In the end, what you'll be is both isolated and alienated from your human peers, and that's just not where anyone should end up.

6. The same again, but with romance

Oh god, I really struggle to read all those pieces about people who've fallen in love with the AI. I find it incredibly tough to have to reckon with the fact that there are quite a few people out there who romantically jack off to something that doesn't have a human at the end of it, and merely sounds, if you squint, like maybe it does. It makes me feel quite viscerally sad and uncomfortable.

Now, this may be the point at which you'd like to pipe up and say "well, Marie, not everyone is lucky in love, and being smitten with the computer is maybe better than not being smitten at all", and I would like to say: pah! I know what I'm talking about! I bet I've spent more time being single than you have!

I was never one of those kids who had relationships at school, and I had a string of short, questionable relationships from around 18 til 23, then I had nothing until I was 28. Nothing. Well, I did have the occasional dalliance with someone usually underwhelming and unsuitable but, for five entire years in my twenties, I did not have anything that even vaguely resembled a short-term relationship, let alone a long-term one. Five entire years! At the stage of life when more or less everyone else was busy getting coupled up!

For the avoidance of doubt: it sucked. It sucked so bad. The crushing loneliness was frequently devastating. What happened eventually, though, is that I decided I had to do something about it. I quit smoking; started exercising; ate healthier food; cut down on my drinking. In the end, my love life did end up coming back from the dead but, even if it hadn't, those life changes would have still remained positive ones.

Similarly, this dry half-decade made me read more, go to the cinema a lot, go to exhibitions, go for walks, take pictures with my film camera. It forced me to keep myself busy, because I knew I was the only person I could rely on. That was a pretty hefty silver lining, if you ask me. What would have happened if I'd decided to just swoon at the screen instead, endlessly messaging a bot which indulged my desires? I just don't think I would have done any of those good things. I wouldn't have had any meaningful impetus to change my life. What a missed opportunity that would have been.

7. If not you, then someone else

You may have read the last few entries and thought "hmmm, but that wouldn't happen to me though, or to anyone I know" to which I would say: good for you! I don't think it would happen to me either! Or to most of my friends! But that doesn't mean it wouldn't happen at all. I have people I love dearly in my life who suffer from, to pick one example, bipolar disorder, and there's no telling what would happen to them if "mania" and "going on chatGPT" ever end up occurring at the same time.

There are a lot of mentally ill and vulnerable and not very bright and quite broken people in this world, and it just isn't clear to me that any of the GenAI companies have put any thought into what their products would do to all those people. That's why it's important that the media keeps covering those quite edge cases, of people using ChatGPT so much that it ends up destroying their lives.

It's absolutely true that the majority of people won't be as unstable, dim or suggestible as them, but that's not the point. If, say, OpenAI doesn't prove that it's trying very, very hard to mitigate potential risks to those people, then those features stop being anecdotes, and start being vital internal policy stories. It doesn't really matter that you or I wouldn't do X or Y, in the same way that a lot of anti-fraud policies aren't aimed at the tech literate, but at people who make for easy marks.

Being in some way vulnerable isn't a crime, but it's also something that cannot be ignored by powerful people, yet that currently is exactly what's happening.

8. The confidence trick

On a related note - I just hate how confident sounding GenAI is. You google something and on the page, right there, is something that is factually incorrect, but which the page tells you about like it has never been more certain of anything ever. Why is that? Would it be that hard for whoever built those machines to add in some sort of tone moderator, whereby a definite fact is discussed assertively, but a "eh, maybe" sounds like just that?

I have no idea if it's possible to build this but - crucially - it's none of my problem. It drives me insane that so many people have accepted this thing into their lives, which brazenly lies to them on a daily basis. I don't understand why and how it's not completely eroding their sense of trust. If you spend your life being told, confidently, about things you know you should double-check as they may be completely wrong, what impact does that end up having on the way you interact with the world?

The "hallucinations" are a huge problem, obviously but I do think the assertiveness is the biggest, reddest flag for me. At least some of my hostility would probably become less strident if GenAI sounded unsure more often.

9. The opposite of blogs

One of my favourite quotes in Escape, my book on the internet, came from Tea Hacic-Vlahovic, a blogger turned author I've been following online for over 15 years. I can't really be bothered digging it out now but her point was: old-timey bloggers and today's influencers are entirely different, because the former would live exciting lives then go home and write about them on the computer, while the latter do exciting things for the very specific purpose of showcasing them on the internet. The generational shift was subtle but fundamental.

Similarly, I feel quite strongly that our horizontal, DIY internet's purpose was to bring people together, and today's is to keep them apart. Small-scale blogs and forums were about meeting like-minded dweebs, and forming new friendships unconstrained by age or geography. Today's social media platforms often divide us into creators and consumers; people whose job it is to go after others' eyeballs, and people whose attention is the product. It's more hierarchical, financially driven, and full of slop.

In what world is AI going to make that better? Already, I've caught myself reading stuff online and wondering if it'd been written by a real human, or merely spat out by a GenAI tool fed a series of prompts. I find it soul-destroying, quite literally: at risk of sounding like a tedious hippie, I do believe that one of the main purposes of being alive is to form connections with others, in any way we can. Getting a robot to write words for you means building an extra layer between you and whoever you see as your audience.

It builds a distance between people, which all feels sad and ironic as the old internet was so good at removing so many of the layers which meant that like-minded people couldn't have met in real life. It's like petting a dog while wearing gloves: you're technically touching but, without feeling that soft and warm fur on your fingertips, so much of the joy you could have had remains just out of reach.

10. Not every problem has a solution

People die. People die! I should know, I lost five of them in the space of around two years recently. It was horrible. Some of them were old and others were young; some deaths we saw coming, others took us by surprise. I went, in quite a short period of time, from being someone who thankfully had spent little time thinking about grief, to one who now greets it as a dear friend. I really wouldn't recommend it.

I also don't have all the answers on what grief does to you, because it's such a personal, ever-changing topic. I don't even really know how I feel about grief myself. What I do know, though, is that I would rather stab myself in the eyeball than chat away with the AI version of any of those people. I just cannot conceive of a world in which it would help. Grief means getting used to the idea of life being finite, and it's pretty impossible to do as it is, so I can't imagine what a mindfuck it would be if you were able to occasionally crack and go chat with your dead grandad on the phone.

More broadly, I really bristle at the idea that Technology Can Solve Everything. I don't think it can. I reject that premise entirely. Still, you can see it everywhere these days. Creepy and gross AI-generated "nudes" are part of that same spectrum, I think. The guys who make them just can't accept that there's someone they want to see naked, but won't ever get to see naked.

It is, in my opinion, extremely bad and daft to give them the message that, actually, with the right prompts and collection of pre-existing pictures, they can get a version of the tits they've been thinking of for so long. It's obviously hugely violating to the people being pictured - christ, that goes without saying - but it's also harmful to the people doing the generating.

A crucial part of life is making peace with not getting everything you want. I completely - and I mean, completely - suck at drawing and, maybe, if I really keep going, and I get lucky, I'll get vaguely decent at oil painting. That'll be the height of my artistic achievement. I'll never be a famous artist. That's fine! You can't always get what you want. Everyone dies. Not everyone wants to show you their genitals. I'll never win the Turner Prize. We all have our crosses to bear. It's what brings us together. We shouldn't listen to anyone who tells us otherwise.

11. God, look at them

I started with an obvious point and I want to finish with another obvious point. I hate the people who are building generative AI and trying to sell it to us. They're the same morons who tried to convince us that NFTs and the metaverse were the future, and they were wrong on both counts. Why am I meant to trust them now? I don't want to trust them. So many of them have bent the knee to Trump. So many of them clearly have no interest in women, minorities, or anyone who doesn't look or sound like them.

Also: so many of them are visibly quite mad, and not even in a fun way. Their auras are all out of whack. I find it impossible to look at them and think "Oh yeah sure, have all my personal data! By all means, I'm happy to outsource my critical reasoning to your products!". There's maybe a world in which GenAI, touted by other people, would have left me feeling less grossed out, but it is: not this one. Fucking freaks, the lot of them.

I would add, as a final note, that maybe they can tell that whatever it is they're building isn't actually all it's cracked up to be. That's why they're forcing it on us like we're one collective toddler refusing to eat his mashed potatoes. They're jamming the spoon in our mouth, again and again, without stopping to ask themselves if that's what we really want.

AI is everywhere now: in our phones, our search engines, our social media platforms, our pieces of software. It's both unreliable and unavoidable. We've not been given a choice at all, and I guess I find it hard to accept that something is going to change my life just because someone else said so.

Build something good, show us how it works, explain to us why we ought to be using it, and perhaps - perhaps! - I'll then decide to trust you, and get in on the action. Don't just jam that plastic spoon full of mash against my teeth and expect me not to wail. I will wail, believe me.