Some Quick-and-Dirty Thoughts on AI's Future - awful.systems
awful.systems
(This is basically an expanded version of a comment on the weekly Stubsack - I’ve linked it above for convenience’s sake.) This is pure gut instinct, but I’m starting to get the feeling this AI bubble’s gonna destroy the concept of artificial intelligence as we know it. On the artistic front, there’s the general tidal wave of AI-generated slop [https://nymag.com/intelligencer/article/ai-generated-content-internet-online-slop-spam.html] (which I’ve come to term “the slop-nami”) which has come to drown the Internet in zero-effort garbage, interesting only when the art’s utterly insane [https://twitter.com/FacebookAIslop] or its prompter gets publicly humiliated [https://twitter.com/venturetwins/status/1844127816612499907], and, to quote Line Goes Up, “derivative, lazy, ugly, hollow, and boring” the other 99% of the time. (And all while the AI industry steals artists’ work [https://spectrum.ieee.org/midjourney-copyright], destroys their livelihoods [https://archive.is/GGNoC] and shamelessly mocks their victims throughout [https://twitter.com/tsarnick/status/1803920566761722166].) On the “intelligence” front, the bubble’s given us public and spectacular failures of reasoning/logic like Google gluing pizza [https://www.theverge.com/2024/6/11/24176490/mm-delicious-glue] and eating onions [https://www.avclub.com/google-s-ai-feeds-answers-from-the-onion-1851500362], ChatGPT sucking at chess [https://www.chess.com/article/view/chatgpt-chess-advice] and briefly losing its shit [https://arstechnica.com/information-technology/2024/02/chatgpt-alarms-users-by-spitting-out-shakespearean-nonsense-and-rambling/], and so much more - even in the absence of formal proof LLMs can’t reason [https://garymarcus.substack.com/p/llms-dont-do-formal-reasoning-and], its not hard to conclude they’re far from intelligent. All of this is, of course, happening whilst the tech industry as a whole is hyping the ever-loving FUCK out of AI, breathlessly praising its supposed creativity/intelligence/brilliance and relentlessly claiming that they’re on the cusp of AGI/superintelligence/whatever-the-fuck-they’re-calling-it-right-now, they just need to raise a few more billion dollars and boil a few more hundred lakes [https://www.standard.co.uk/news/tech/ai-chatgpt-water-power-usage-b1106592.html] and kill a few more hundred species [https://pivot-to-ai.com/2024/07/23/data-centers-risk-missing-us-climate-goals-especially-with-ai/] and enable a few more months of SEO [https://wired.me/gear/google-search-ai-spam/] and scams [https://www.abc.net.au/news/2023-04-12/artificial-intelligence-ai-scams-voice-cloning-phishing-chatgpt/102064086] and spam [https://www.businessinsider.com/ai-spam-google-ruin-internet-search-scams-chatgpt-2024-1] and slop [https://futurism.com/the-byte/source-ai-slop-facebook] and soulless shameless scum-sucking shitbags senselessly shitting over everything that was good about the Internet. ---- The public’s collective consciousness was ready for a lot of futures regarding AI - a future where it took everyone’s jobs, a future where it started the apocalypse, a future where it brought about utopia, etcetera. A future where AI ruins everything by being utterly, fundamentally incompetent, like the one we’re living in now? That’s a future the public was not ready for - sci-fi writers weren’t playing much the idea of “incompetent AI ruins everything” (Paranoia [https://en.wikipedia.org/wiki/Paranoia_(role-playing_game)] is the only example I know of), and the tech press wasn’t gonna run stories about AI’s faults until it became unignorable (like that lawyer who got in trouble for taking ChatGPT at its word [https://arstechnica.com/tech-policy/2023/05/lawyer-cited-6-fake-cases-made-up-by-chatgpt-judge-calls-it-unprecedented/]). Now, of course, the public’s had plenty of time to let the reality of this current AI bubble sink in, to watch as the AI industry tries and fails to fix the unfixable hallucination issue, to watch the likes of CrAIyon and Midjourney continually fail to produce anything even remotely worth the effort of typing out a prompt, to watch AI creep into and enshittify every waking aspect of their lives as their bosses and higher-ups buy the hype hook, line and fucking sinker. ---- All this, I feel, has built an image of AI as inherently incapable of humanlike intelligence/creativity (let alone Superintelligencetm), no matter how many server farms you build or oceans of water you boil. Especially so on the creativity front - publicly rejecting AI, like what Procreate [https://twitter.com/Procreate/status/1825311104584802470] and Schoolism [https://twitter.com/SchoolismLIVE/status/1843796461638824034] did, earns you an instant standing ovation, whilst openly shilling it (like PC Gamer [https://twitter.com/pcgamer/status/1844796247044948363] or The Bookseller [https://twitter.com/thebookseller/status/1841828201804300491]) or showcasing it (like Justine Moore [https://twitter.com/venturetwins/status/1844127816612499907], Proper Prompter [https://twitter.com/ProperPrompter/status/1809206091764613487] or Luma Labs [https://twitter.com/LumaLabsAI/status/1800921393321934915]) gets you publicly and relentlessly lambasted. To quote Baldur Bjarnason [https://www.baldurbjarnason.com/2024/slop-framing-failure-as-success/], the “E-number additive, but for creative work” connotation of “AI” is more-or-less a permanent fixture in the public’s mind. I don’t have any pithy quote to wrap this up, but to take a shot in the dark, I expect we’re gonna see a particularly long and harsh AI winter once the bubble bursts - one fueled not only by disappointment in the failures of LLMs, but widespread public outrage at the massive damage the bubble inflicted, with AI funding facing heavy scrutiny as the public comes to treat any research into the field as done with potentally malicious intent.

(No clue why I didn’t get around to this earlier, I’ve had this in drafts for too long.)

Eight months ago, as you probably know, I predicted the current AI bubble would destroy artificial intelligence as a concept, focusing on the unrelenting slop and failures of AI, and on the near-universal backlash it receives whenever it rears its ugly, slop-ridden head.

As it turns out, I had completely failed to recognise the the political elements of this entire bubble. In retrospect, I should’ve recognised it a lot fucking earlier.

Between Baldur Bjarnason outlining the esoteric fascist elements at the heart of the AI bubble, AI slop’s enthusiastic adoption by fascists of all stripes, Damien Williams’ notes on authoritarians’ love for gen-AI, Ashley Lynch calling AI slop inherently fascist, its become clear the outright fascist nature of AI has been staring me in the face the entire time.

With all that in mind, I’d like to expand on my previous piece with three additional predictions.

  1. AI-as-Fascism

Right off the bat, I expect AI as a concept will pick up a public perception of being inherently fascist, or at least a tool of fascism. Beyond all the ink spilled about AI’s fascist nature, Donald Trump going all-in on AI has done plenty to link his administration with the tech, whether through making AI slop of deportees, or letting Elon Musk’s AI Poweredtm Department Of Government Efficiency go to town on the federal gov.

Long-term, I expect this will hamper future attempts to start new AI bubbles/AI springs, as attempts to revive the tech get treated as morally equivalent to creating the Fourth Reich.

  1. The Wider Tech Industry

On a wider front, I expect the tech industry at large will pick up a similar stench of Eau de Fash as well. Whilst the tech industry has long enjoyed a perception of apoliticality (which James Allen-Robertson has talked about (spoilers for Devs BTW), their own heavy involvement with the Trump administration has done plenty to undermine that.

Between their sucking up to Trump and AI’s own stench of Eau de Fash, I can see the public starting to view the tech industry at large as a Nazi bar once the bubble bursts. Silicon Valley’s given 'em plenty of reason to do so.

  1. A Bone for The Humanities

Ending this off on a vaguely positive note, I suspect the bubble’s burst will earn the humanities some begrudging respect once the dust settles - primarily through cannibalising a fair bit of the cultural cachet that STEM has built up over the decades.

On one front, the slop-nami has given us an absolute torrent of slop flooding the Internet, notable both in its uniquely inhuman shittiness, and in AI bros’ breathless adoration of it. Given that, I expect programmers/software engineers will come to be viewed as inherently incapable of making anything on par with anyone who has taken up art as a hobby/profession, and incapable of understanding art with any sort of depth to it.

Additionally, I suspect a stereotype of programmers/software engineers being hostile to art/artists may form, thanks to the rather drastic toll this bubble has had on artists, and the ongoing rhetoric of “democratising art” that the bubble’s given us. (The age-old “learn to code” adage may also come back to haunt them as well, if this comes to pass.)

On a wider front, the breathless “AI doomsday” criti-hype, more general over-the-top AI hype, and nonstop hallucination-induced mishaps will likely also contribute to making STEM as a discipline look out-of-touch with reality, making the humanities look grounded and reasonable by comparison as the public looks in confusion at AI bros’ inability to recognise LLMs’ shittiness for what it is.