I want to talk a little about AI.
I'm not a doomer. I don't believe it'll be the end of humanity. I don't think it will have as big an impact on the workforce as we think it will. But I don't like it, either. LLMs and other kinds of generative statistical models are really interesting techincal achievments; who would've thought that a denoising algorithm could resolve a textual description to a jpeg? Wild.
Generative models are symptomatic of a great accelerating decline though. We've become hostile to subsitence. We've glorified grifter culture and embedded get rich quick mentalities into multiple hopeless generations - if you don't have a seed round before 25, you're doomed to obscurity (do not get me started on ycombinator). Thankfully there is some strong resistence emerging to this; though I fear not strong enough.
Generative models come with a promise - you, even you, yes you, who thought you were doomed, can be among the ranks of bezos and zuckerberg (fucking gross???) using our tools to do the work for you. The "It sells itself" of MLM schemes comes to mind. You just need to believe in our model and it will deliver - you don't need an idea, you can just ask for one, spend $25/mo on cursor and tell it to build it. The buzzwords of "Conversions", "B2C Sales" and "Retention" come to mind. The worst people imaginable can cut out the middlepeople of developers, designers and testers and directly sell - let's be honest - their desire to be megarich, packaged as a digital product and shipped to millions. They can call themselves "founders" and wait to be bought out by Atlassian, who would obviously come to them for the paying userbase and leads that their app gathered rather than do exactly the same that they did, and generate their own. The big red "Make Game Button" GarageGames never delivered is here at last!
There is an argument that generative models have democracised creation and made it more accessible; I don't buy it. Creation has never been inaccessible to the same people with the capabilities to use generative models; the process is just incompatible with get rich quick grifting. If I can make a parallel; you've read Frankenstein right? Great novel - considered by many one of the first published science fiction stories and written by Mary Shelley as a way to repel Lord Byron. It is subtitled 'The Modern Prometheus' - Prometheus being a greek myth, about a Titan who stole fire from the gods to bring it to man and in a sense, bringing man and himself closer to god. Victor Frankenstein, similarly, created a man--out of the flesh of his deceased contemporaries. He sought to defy god and prove that he himself approached their greatness, that he too could create life. But he was not defying god. Humans are perfectly capable of creating life; we birth new humans every second of every day (greater than the replacement rate of humans dying, even). In defying god, Vic was in fact emulating women OH SHIT FRANKENSTEIN IS A TRANS ALLEGOR-
All humans are already, innately capable of creation. Crafts are skills that can be honed - whether you're painting, scultping, composing a symphony - these are all within each of our grasps. But they take time. They take understanding. And in Victor's case, it takes talking to a woman. We often mistake our hubris for greatness in potentia and this is one of our downfalls as humans, I mean, how many of us have tried to start a hobby or learn a skill and put it down when we weren't good at it straight away? I did that with graphics programming. Generative models come with a promise to be able to skip all that - the grueling process of personal development, the weeks and months of practice and learning. They skip you to the end product. And that is a temptation that is very, very hard to resist. If we could load a Matrix training program into our brains and become world-class Kung-Fu masters, why the hell would we not? I've been trying to learn Japanese. As a language it is almost incompatible with everything I know in my native tongue - but slowly, I'm grokking more every week. But when you learn a new language, you can feel parts of your brain changing to accomodate a different way of understanding and expressing. If I could just read and speak the language without these changes happening, instantly, I would miss out on the joy of actually feeling and seeing that growth. I wouldn't change as a person, other than that I could hold a conversation with the tiny punk bar owners in Osaka better. But you know what it would be great for? Employment. I could make so much money launching apps and working wth companies wanting to got to market in Japan without spending any of my precious time or money. And my employer could exploit that skill to generate even more money for themselves.
This applies as much to art as it does to code. The commercialisation of art has been a debated topic for decades and I'm not going to get into it here. What is the purpose of art in a world where the only motivation is conversions and sales? We trade the process of art for the profit of art; time is money, after all. I don't know if I want to live in a world where it took me longer to read a novel, or an article, or an analysis, than it did to write it. There are very few meals that take longer to eat than prepare, and we don't consider them viable to live on (as much as kraft mac n cheese can hit the spot, it's nice to eat a vegetable once in a while). If we don't accept food that isn't viable to live on nutritionally, why would we settle for art that isn't viable to live on emotionally?
The thing that suffers most here is the joy of creation. When I started learning to code, I didn't have it as a career in mind. I wasn't even a tween and I wanted to make custom worlds in ZZT and Commander Keen (and if an eight year old can learn C and x86 assembly in 2000, you can learn F# now). I'm a musician, I never picked up the violin of the guitar thinking I'd be a concert master, or a rock star. I just wanted to make things. I had feelings that I had to excise out of me. I had beauty in my head I wanted to show the world, even if the world was three friends and my parents. We can now algorthmically determine what music is liekly to be placed on auto-generated spotify playlists to generate the most income per dollar spent on its production. We can detemine how movies will perform in cinemas and on streaming services before they're filmed. In programing circles, there is a saying: "Premature optimisation is the root of all evil". We have optimised the hell out of art. Its not that in a few years our soundtrack and TV screens will be overwhelmed by the equivelant of hotel room art. It's that they already are, and it's still not enough.
In the 2010s there was a renaissance of creation. We had makerspaces. Arduino produced a low point of entry for people wanting to explore hardware design, Raspberry Pi helped people built their own computers from a single board, and people were buying 3D printers in droves because they wanted to make cool shit. We still have those things - but it's still hard. You still need to learn how circuits work, what 'registers' mean, how to structure an STL so it will print properly. But they are still low-cost, low barrier of entry tools you can hammer again and again to figure out how to build what you have in our head and show the world. In the 2000s the Flip video camera came out. It was revolutionary for students - you could press a button and make your own short film. We had decades of truly democratising some of the historically most inaccessible modes of creation and we had people, creative young minds, discovering and rediscovering the joy of creation. We had MySpace letting bedroom musicians show their songs to the world at an unprecedented scale. Bandcamp and SoundCloud made it even easier, and gave them ways to sell their merch and music to more people than ever before by completely bypassing the predatory record label model.
But now we're in a world where the property market is making us homeless. People who work long, grueling hours cannot afford to eat. People are not having children because they cannot afford to feed and house themselves, let alone any dependants. There is war, poverty and abject cruelty surrounding every one of us if we see it or not. Climate change is hitting us in the place a lot of people weren't told about: our wallets. We're desperate and uninsurable. The 2020's stared with a message - "We're locked down. You have all the time in the world. If you're not rich, smart and successful by the time we're out, you're a lazy parasite". And so the grifts started. Web3, Crypto, NFTs, and endless list of get rich quick ponzi schemes.
People want to believe that generative models will be their saviour. I cannot blame them. Subsistence is harder than it has ever been for more people than it ever has been. In previous generations where subsitence has been easly obtainable to the masses, it increasingly is not at all. In a world that is hostile to subsistence, generative models are just another symptom of much more fundamental problems with the world we are no closer to solving. And generative models won't solve them either.
But they can show you an astronaut riding a rainbow unicorn in space. That's something, right?