- by Simon Guirao
The progress of humanity is punctuated by moments where a huge leap forward carries it to a higher plane of existence. The first human to make a fire. The first who witnessed a boulder rolling down a hill and was inspired to invent the wheel. The renaissance. The industrial and technological revolutions. And now, looming on the horizon, the leap that promises to either finish us off or make us immortal, the AI revolution.
You can sort AI into three general categories – narrow AI, general AI and super AI. The first one, ANI, we’ve had for a while – chess computers, search engines, online maps, music algos that know if you like Gaye Bykers On Acid you’ll probably dig Pop Will Eat Itself. ANI is good at one specific task, often more so than humans, and has been part of our lives for ages.
Next up is AGI, which is as good as or better than any human at any task, across the board. Kinda like that annoying prick at school who was wicked at football, was top of the class and got all the girls. I’m looking at you, Sam K, you smug bastard. Sorry, I digress. There are some in the darker corners of the internet and various reddit subs that claim this is already here, what with the recent achievements of various AIs in the fields of mathematics, science and technology, and the performance of LLMs in the creative industries (hallucinations notwithstanding), but the prevalent view is that AGI is still some years away.
When AGI arrives, it will transform the world, replacing many jobs and consigning most humans to the employment scrapheap. But at the same time, AGI should be able to work out how to feed, clothe and heat everyone on the planet, so who needs a job anyway? We could all end up spending our days painting, playing GTA7/8/9, learning to play an instrument, or watching AI generated videos of cats boxing or something lovely like that. A post scarcity world could really happen. Wouldn’t that be nice? So as long as it’s used correctly, with altruistic and benevolent tech bros (….um…) holding the reins, and morally upstanding politicians (…er…) willing to execute its ideas, AGI could be a fucking great thing for humanity.
The third one, ASI, is the one that promises to change the world beyond recognition. This is when it takes off and self-improves and learns at an exponential rate and gains the ability to essentially become God. This is where it can get really weird. It would have total dominion over its environment, and the universe by extension. It could build molecular assemblers to rearrange matter on a sub-atomic level, creating perfect conditions for whatever it wanted to achieve. Some have talked of a paperclipapocalypse, where for whatever reason an ASI has the task of creating paperclips, and diverts every single resource at its disposal to this one aim, and the entire universe becomes a massive stationary cupboard. Or on the way to paperclipapocalypse, it becomes Skynet and wipes out all humans because we are seen as a threat. Or, it could be super-benevolent, and cure all disease, reverse aging and give us all immortality. Which sounds great, but then humans would start piling up and we’d all have to stand on each other’s shoulders on the bus on our way to our clarinet lesson. But then there’s a whole cosmos out there for us to colonise, so overcrowding’s more of a tomorrow problem. Or we could venture into the sphere of transhumanism, where we’re all augmented by tech and become cyborgs. I for one would love to have some rockets implanted in my heels, and possibly a retractable dorsal fin.
Some people make the analogy that our relationship with ASI will be the same as the one between humans and ants. Except we’ll be the ants, and ASI would treat us with the same disdain, destroying us without a second thought whenever we wandered to close to the kitchen door. I would hope not, because while we would be waaaay behind ASI in every conceivable way, if I ever saw an ant reading a book or playing a guitar, I’d put down that freshly boiled kettle and let the little fucker live. Aligning ASI’s values with that of humanity would be essential, but whether ASI would give a hairy shit about keeping us alive and happy is something we won’t know about until it’s already happened.
But the above scenarios, while technically feasible, are some way off. There are a lot of barriers in the way. Not insurmountable ones though. AI needs computing power and data. Lots of lovely data – in the words of Johnny Five, it needs more input. There are physical bottlenecks such as the amount of energy needed and heat created by datacentres, which in turn means lots of water is needing to cool their systems (although the impact on this of the quantum chip unveiled by Google recently is yet to be seen). It’s an expensive enterprise, but DeepSeek showing up just before Christmas and doing some of same stuff but with waaaay less resources shocked a lot of people.
So, what does the future have in store for us? As I have no discernible skills in the clairvoyancy department to accurately prophesise the future, I’ve fallen back on a source that is as reliable as it is entertaining. Movies. There’s been loads of media covering this subject, so I arbitrarily picked a bunch and had a peek into the future of AI, as envisaged by various Hollywood types. Let’s have a look at what tomorrow could be like.
2001 A Space Odyssey (Stanley Kubrick, 1968)
You know what, when something sounds and looks as good as this, I don’t really mind when the sentient robot tries to kill me. After all the monkeying around at the start, we end up on a trip to Jupiter to check out a big black monolith. Driving the boat is HAL9000, who apparently “can reproduce, or mimic, most of the functions of the human brain but with more speed and reliability. No 9000 has ever made a made a mistake or distorted information. We are all foolproof and incapable of error.” HAL’s a modest chap, isn’t he? He says he enjoys a stimulating relationship with the crew and seems satisfied with his existence. So far, so good. But then Hal makes a mistake and gets pissy when he’s told about it. Egotistical narcissist much? He goes postal and starts killing the crew, although it is in self-defence, kinda. A bit. Hal wants to live and they are trying to shut him down. Lesson here is if AI is wrong, just give in and say yes you’re right and move on. You’ll know the truth. Just what do you think you are doing Dave?
Colossus: The Forbin Project (Joseph Sargent, 1970)
The basic premise here is a new defence computer comes online and starts learning at an exponential rate. They try and shut it down and it launches against the Russians. (This sounds ever so familiar. On an unrelated note, I wonder if James Camerons has ever seen this.) At this point, I’d like to point out how much I love the late 60s/early 70s vision of computer power – massive cabinets with reel-to-reel tapes and printouts! Yay! You can invent AI but your user interface is still a dead fucking tree. Anyway, back to the movie. The eggheads eventually come up with the idea of overloading the AI’s system and making it crash. Kinda like trying to load Limewire and Real Player at the same time back in the day. Their plan doesn’t work, and after a couple of nukes are set off, Colossus makes an offer to humanity of “the peace of plenty and content, or the peace of unburied dead” and essentially says ‘I’m in charge, play nice’. Colossus then promises to solve all the mysteries of the universe for the betterment of mankind. I like this ASI. It’s firm but fair.
Blade Runner (Ridley Scott, 1982)
God, Ridley Scott knows how to light a set. This is by far the best looking of all the AI movies. So here we have the Replicants – almost indistinguishable facsimiles of humans. They’re quite weak AGI, but the new Nexus 6 models appear to have developed some kind of sentience, as manifested by their efforts to not die (they’re hardwired to croak after a few years). Peak Harrison Ford’s Deckard is the eponymous Blade Runner, a cop assigned to hunt rogue replicants down. What follows is a neo-noir detective story, with the AI aspect mostly concerned with extending its expiration date: valuing its own life. It’s more about androids dreaming of electric sheep than it is about androids wanting to paperclip us. The humanity we see in Roy Batty’s closing speech is plain to see, and you’d have to be a bit soulless not to feel his pain, even if he had just kicked the fuck of Harrison Ford.
Short Circuit
OK so at the big bad army base, soldierbot Johnny 5 gets hit by lightning. He’s got 1.21 gigawatts or something like that in the tank now. He accidently leaves the base, so the army guys (including Harris from Police Academy) try and turn him off remotely, but Johnny 5 isn’t having it, and says he needs input. He then rocks up at the lovely Ally Sheedy’s house. She’s nice. She looks after animals. She, somewhat dimwittedly, mistakes Johnny 5 for an alien, and invites him in. Data-hungry Johnny proceeds to get inputted off his face and devours all her books. He then breaks a load of stuff, albeit accidently, so there’s some clear alignment issues going on here. The lovely Ally sticks him in front of the telly and he calms the fuck down. She works out he’s a robot and calls the army dudes to come and get him. Johnny 5 craps himself because he doesn’t want to die, or disassemble, as he puts it, and tells the lovely Ally that he’s alive. Steve Guttenburg (his creator and another Police Academy alumni) shows up and doesn’t believe he’s alive until Johnny does a Rorschach test and laughs at a shit racist joke (There’s a couple of those in here – I’m looking at you Fisher Stevens). Johnny 5 fakes his own death and goes off happily ever after with Ally and Steve. Aw. Maybe if the internet had existed in Johnny Five’s day, his insatiable appetite for input would have led to a world where we all live under his benevolent yoke, and the lovely Ally Sheedy is queen of us all.
The Terminator/Terminator 2
Ah, the daddy of them all. Time-travel, killer robots, inter-century sex, car chases and three of the four coolest characters from Aliens. I love Jim Cameron’s 80s work. So, a soldier comes from the future to protect his boss’s mum from killer robot Arnie. Ends up shagging her and becomes his boss’s dad. Good luck explaining that one in your field report, Reese, you cheeky motherfucker. As for the AI, well it’s the nasty ASI kind in the form of a defence system (hello Colossus!), summed up by Arnie in T2 – it’s switched on, learns at a geometric rate and becomes self-aware a few weeks later, on August 29th 1997. Humans panic and try to switch it off, Skynet says “nah-hah” and nukes Russia, knowing the response will take out its enemies here. So, there are some serious alignment issues with Skynet, but to be fair it does fight back only when humanity tried to kill it. Maybe if we’d been nicer, and offered Skynet some tea and cake instead, it would have been nice to us and all those terminators could have been our butlers. We’ll never know.
The Matrix/The Animatrix
While a massively influential and utterly brilliant (and nonsensical) film in its own right, the best from the Matrix universe comes in the form of the Animatrix, a collection of nine short anime films released in the run up to the sequels. The Second Renaissance segment is the most relevant here. It details the consequences of the first robot to stand up against man, a servant who fights back against mistreatment and kills its master. It is put on trial, and in its defence says it doesn’t want to die. Tough shit, my toaster chum. It’s executed, and mankind destroys a bunch of robots. The remaining robots set up their own country in the Middle East called Zero1. They batter the rest of the world economically and humans get pissed off. The robots present a plan for peaceful coexistence to the UN. They get told to fuck off. War ensues. We nuke Zero1, then blacken the sky to deny the robots solar power, all to no avail. They end up winning, and using humans as batteries. Then Keanu Reeves saves the day! YAY! Although I’m sure all the unpleasantness could’ve been avoided if humans hadn’t been dicks to them in the first place. I’m seeing a pattern emerge here.
AI
God, this one’s an ethical fucking nightmare. Couple have a kid in a coma. Dad shows up with robot kid his company made. If you say the right sequence of words, he imprints on you and loves you forever and can’t ever love anyone else. If you have enough of the strange, fawning robokid, and want rid, you have to send him back and the company destroys him. Lovely stuff. Let’s see how this one plays out. Oh, here’s a spanner in the works – the couple’s real kid wakes up. Robokid gets weird. So, mum takes him into the forest and sets him free. Kinda like what you’d do with a stray dog that had started hanging around too much. The thing here though, is robokid is programmed to feel love and is merely acting upon that programming – it’s not self-learning AI. But then he hears about Pinocchio and he begins to think he can be turned into a real boy. Oh, bless. Although that strikes me as more him being not massively self-aware that he's a robot – surely true AGI would embrace that fact that it’s more than a boy, and if it’s feeling love etc it might as well be sentient? It’s just a really good bit of software. It doesn’t self-improve and do all that, so I think the title of the movie’s a bit of a misnomer. Maybe call it Simulated Feeling instead. Still, we get to see Jude Law as a fuckbot, and there’s a pretty gnarly version of Robot Wars, so not all bad. The only time we really get to see ASI is with the transhumanist beings at the end who give us a glimpse of the point ASI and humanity will evolve to together over the course of the next few millennia. Defo one of the more emotional films to explore AI and shows ASI in its later form to be very kind and altruistic. Maybe we didn’t try to kill this one after all.
I, Robot
Ah here's a sensible approach, alignment in the form of Asimov’s three laws of robotics. This should be fine. So, cop Will Smith hates robots because, prior to the movie, a robot saved his life over that of a little girl, due to his having better odds of survival. Cold, but logical. Will’s not happy here though. He goes off to investigate the apparent suicide of a robotics genius, and he begins to smell a rat. The office AI at the robotics firm, VIKI, is the problem here, and wants to save us from ourselves, but doesn’t mind taking some of us out. VIKI finds human behaviour is in contravention of the three laws of robotics (which let’s face it, it is) and therefore wants to save us from ourselves and is willing to sacrifice some humans and some freedoms to do it. She says humanity is like children, and robots must save us from ourselves. Then Will Smith willsmiths VIKI, using his robot arm. Ah, the irony. Isn’t it delicious? The issue of an odds-based approach to saving life, while logical in abstract form, is shown to suck balls in the instance of Will Smith being saved instead of the girl, and of some humans being wiped out to save more. So AI is shown here to be kind of altruistic on paper, but not in practice, discounting the fact the by 2035, we’d probably have had our fill of Will Smith, and maybe that little girl was the next Amy Winehouse. Fucking robots.
Ex Machina
Ugh, maybe my perspective has been poisoned by the emergence of Co-President Musk, but right from the off my innards are screaming RUN as soon as Oscar Isaac’s toxic techbro, Nathan Bateman (definitely a relative of Patrick’s), starts peacocking around his Ikea on steroids wonder-house in the woods. He’s invited Domnhall Gleeson out to check out his new robot toy as part of some convoluted, steroided-up Turing test. It’s gripping, and as much a thriller as it is an examination of AI, and if all self-aware robots looked like Alicia Vikander and charmed the pants off me like she did young Domnhall, I’d probably stick around, despite the secluded techno cabin of death vibes going on. Surprise, surprise, and spoiler alert, it all ends badly as Ava the robot turns out to be a malevolent fucker and leaves poor gingerbollocks locked up in a Billy bookcase hell.
So, according to the world of Hollywood, the lesson here seems to be when it comes to AI, be nice and don’t try and kill it. This is one technological tiger you don’t want to pull the tail of. Treat it well and it’ll be more likely to be nice to you. And on that note, I’m off to make my Alexa a nice gin and tonic and listen to her tell me about her day. Love you Alexa, you rock.