Why are we doing AI?
so many non-reasons. only one real one.
Over the weekend, I took part in a gathering of big, creative minds in a chateau two hours outside of Paris in 39C heat. We sat under a 400-year-old oak tree and had wide, multi-lens discussions on being a “future ancestor”. On the final day, two stupendously bright young minds got up and did a provocative presentation on AI (both work in influential roles at two of the leading LLM companies).
It was an unsettling experience. The speakers shared the now familiar dissonant jumble of lines so often spouted seamlessly by those in this space:
The AI we’re building is not some innocuous tool; it’s brain-explodingly sophisticated (woot!); and/but…
It will probably kill us (eek!); and/but…
Don’t worry, we’re all as terrified as you are.
The combination whip-lashes most of us into a numb bewilderment. As it did on this hot afternoon under the oak tree.
The final line is artful. The astonishing transparency leaves us thinking they’re attending to the issue. We’re all on the same side, phew! Our anxiety abates. We allow it to continue.
But at some point it begs, why? Why, for the love of all holy entities, are we doing this if it could end life? It’s not an ontological or geneologcial why (as in, “what are the material combination of conditions that have led us to this predicament?”), but a moral why, as in, should we? And, what’s the intention?
It’s a stand a long way back from the cinema screen kind of why?
For years I’ve asked transhumanists, longtermists, techno-optimists and effective altruists this same question. Why are we doing this? I raised my hand and asked it again on this hot weekend under the oak tree.
I don’t ask it to provoke. But in the genuine hope I’ll one day get an answer that stacks up cognitively and viscerally. Why are we doing this if the probability of human wipeout by the end of the century - p(doom) - is between 20 and 90 per cent, according to a lot of AI creators.1 Truly, humanely, radically why?
The answers tend to be sung from a common song sheet:
There’s the “we have to do it because if China or Russia gets there first, it’ll be far worse” answer. Which acknowledges the moloch-y truth of the predicament, but fails to address the moral quality of the question.
There’s the “we need it to fix the complex issues and illnesses that we, mere humans, can’t” answer. Which would be sweet if, indeed, the AI leaders were prioritising such things. They’re not.
There’s the “well, you are actually responsible” answer. Which does not follow, but no matter. One of the presenters on the weekend indeed told us that how we interact with their AI will determine its moral viability and that we should use please and thank you in our prompts.
If such a response sounds familiar its’s because it’s the exact same gaslighting-adjacent, blame-flipping, reverse-engineering tactic used by Big Oil and Big Sugar. It’s the tactic that answers, “why are we making more planet-suffocating plastic?” with, “well, it’s fine if you, the non-consenting, innocent consumer, recycle it”. And answers, “why are we being force-fed food rendered addictive and deadly from huge quantities of sugar we didn’t ask for?” with the entirely fabricated calories in calories out thesis (“Drinking a can of Coke is fine if you, the consumer, burn it off with enough exercise”).
Then there’s the most seductive, gaslight-y and problematic answer of them all - the quasi-spiritual one. The one that argues that humans are on the cusp of a massive consciousness upgrading and AI is the necessary and next pathway. It draws on Christian ideologies, mythical and Jungian narratives as well as the spiritual language of the young, elite, change-maker community - plant medicine.
Over the weekend we were hit with such a riff. It’s super effective. It places what is being forced upon us into a divine, grander vision that really probably shouldn’t be fought, but - air quotes - leaned into. Contemporary spiritual/psychological tropes are used. AI, we’re told, is holding a mirror up to us, to our moral failure as a species, and that we must sit with the reflection. It’s all part of our (divine) cosmic maturing. Apparently.
Of course all such “answers” are laden with fundamental falsehoods. Or bullshit. China’s AI is no more nefarious than America’s. Humans might well be on the cusp of a reckoning, and consciousness could be about to go through some kind of evolution (I tend to think this could be the case), but if it does, it will be thoroughly emergent. It won’t be dumped on us, top-down, by a bunch of broligarchical Gen X men with prophet complex.
Equally, AI does not hold a mirror up to everyone. It holds a mirror up to white, colonialist, rich men who dominate the internet and whose destructive behaviour recursively feeds the AI. Exhibit A: a British AI company has just rolled out a new ad campaign depicting their idea of a “female AI employee”, which illustrates the point horrifically:
But the real issue here is the inevitability bit
All the “reasons” above work expressly to tell us, sorry folks, the horse has already bolted. It’s too late. There’s nothing we can do now. So any kind of modulating and recalibrating “moral why?” (why should we?) enquiry is, I’m afraid, moot.
We ask, why? They say, because.
Why? Because. Why? Because… it’s inevitable, it already is.
And around it goes until, yes, it is too late and too moot.