New Talk – AI FOoM



I was recently asked to give a talk for the DIGITAL TRANSFORMATION & INNOVATION IN BFSI SUMMIT 2024 in Vienna.

I haven’t done one of these things since Covid, so it was useful to update my priors with regard to the future…
…which is bearing down on us like some laser-eyed, fusion-powered Behemoth!

In this talk, I focus on how science fiction can prepare us for AI FOoM! (Fast Onset of Machine intelligence). A scenario that a minority of AI researchers are concerned about, but which has non-trivial consequences for our species.

Aliens!



So, Aliens huh? My gut feeling? All this hysteria is just more of ‘THEIR‘ endless Psyop nonsense to distract us from the Fourth Turning. On the other hand…

I met a drunk guy in the park. He had one of those conference ID tags around his neck. He was sitting on a pallet of Bud Light raving about opportunities in post scarcity deflation. He was clearly making the most of the arbitrage opportunity, so much so, that eventually he staggered off and passed out. The guy was so sloshed he forgot to close his laptop. I had a peak. There was a lot of repetitive energy budgeting and genotype statistical analysis to struggle through but sensing my perplexity, the laptop suggested it could summarize and translate the data into Terran vernacular for aboriginals—

The Vetan, the Quagaars, the Vindaloovians, and the rest of the ten-thousand-strong coalition that make up the GPLC (Galactic Poverty Law Combine) have been visiting Sol3 for a hundred million years.

100M BP – Fish, big lizards, small hairy critters, volcanoes. Recommendation: check back in 50M years.

50M BP – Lots of hairy quadrupeds, some interesting aquatic mega endotherms, significant feathery aerial diversity. Recommendation: check back in 50M years.

1M BP – Bipedal hairy endotherms communicating abstract concepts via sonic telepathy, tool use, marine navigation. Standard technology curve. Recommendation: check back in 1M years.

10K BP – Significant depopulation of mega-fauna. Bipedal endotherms experiencing severe genetic bottlenecking due to comet strikes and a hemisphere-scoped extinction event. Standard technology curve. TTS (time to singularity) 10K years +/- 5K years. Recommendation: distal civilizations may want to start packing for a road-trip.

1K BP – Large-scale social mobilization. Extended cognition […] TTS 1K years +/- 500 years. Recommendation: step up surveillance cadence. Tickets open within the 500LY party zone—start boosting up now, you don’t want to miss this one!

100 years BP – Atomic theory. Branchial Theory. Digital Computation. This is it, galls and zims! The hairy bipeds are going Singularity any decade now! Last chance to book your tickets! This one is going to be a blast!

14th October 2017 – High-density probe insertion despite light sail ejection malfunction. Observation of critical ecosystem degradation. Significant existential self-annihilation risks: current odds are 3:2 against. Evidence of unsanctioned Quagaar technology transfer. Vindaloovian griefing sigils. Genetic diversity has been cached. Post-doc observers will remain on station for the final singularity end-run.

Interesting articles have been written about Sol3. Evolution between major lineages has delivered good papers on Dolphin and Fish convergence. Sonic-telepathy is pretty cool and unusual, remember way back when #GruntingHomo was trending? But you know the deal… there are a hundred billion life-supporting planets in the Galaxy. Dinosaurs are cool, they come in different sizes, but they all taste of chicken. The only thing that properly gins up the Galactic news cycle is some new Sophont species going full-blown Singularity!

This happens rarely; two or three per millennium. All the players like to have delegations on the ground when the fireworks start. Remote civilizations that miss-estimate kick-off time often arrive hundreds or thousands of years early—better a thousand years early than ten minutes late, right?!

We hang around in the Oort Cloud with our press passes around our necks (anatomical analogue) drinking too much (psychotropic analogue), fucking each other (physiological analogue), and killing time. Those of us with a few Singularities under our belt, know there’s always a black market for sneaking down early to get inside scoops or, if you’re from one of the non-compliant factions, negotiating deals before the GPLC lawyers stitch up exclusivity on all the good stuff…

You wouldn’t believe the price of beer down here!

Ooops, the red light just started going ape!
I’m off to put on my Singularity trousers! The fun’s about to start!!
See you on the other side!

What will be left for humanity as the waters of the singularity engulf us?

 

If you see the world as an IQ hierarchy which extends from inanimate to dumb, up through smart, ending up with God, you will be concerned at what is about to happen as our machines make the leap from inert to deity within a single organic human generation. In this world view, we will shortly be deposed from our smug position as the smartest kids in the room, being immediately demoted to something akin to the class hamster or goldfish; entities without agency, whose survival is contingent on the continued indulgence of their custodians.

This might well be the order of things. I, however, a naïvely optimistic, blue-pilled, idealist, 😉  chose to see things otherwise.

Rather than an IQ spectrum or scale, intelligence is a fractal continuum. I can’t go into details here otherwise this short post would become exceedingly long, but you can read some of my thoughts here: What is Consciousness? Essentially, all eukaryotic life is conscious in some sense as it computes its minds into existence through direct interaction with a more fundamental layer of reality.

The reality beneath physics, which computes the physical universe, is thought. Our minds are compartmentalised subunits borrowing awareness from this deeper layer. The layer itself, which I call the ‘Infinite Substrate’ in my books, may experience existence as a single, vast, reality-spanning consciousness. From this perspective, all details are lost beneath blankets of abstraction.

From the perspective of individual consciousnesses, only the relevant details of existence at the scale at which they operate are apparent; all detail below is smothered under Markov blankets of abstraction. The cosmic perspectives of the layers above are likewise incomprehensibly remote.

The ‘Plurality’—my term for Brahman, the ‘Greater Mind’, the Continuum, the Multiverse, etc—is fractal. There will be conscious subunits operating at all scales from cosmic, through stellar, all the way down to the most primitive one-bit wide channel of information which may be a little more than the primal scream of ‘I am therefore I am’.

We are currently not able to create artificial consciousness, which in my conjecture, would require accessing the infinite sea of thought beneath classical physics. We can brute-force something that passes for super intelligence by throwing a whole bunch of atoms and energy at the problem. This will most likely not scale. We WILL have machines that do tasks at a super-intelligent level. This will be achieved by expending country-sized energy budgets to train neural nets on tasks and then cloning them and sticking them in all our gadgets. But these ‘Sages’ as I call them in my books, will not be able to learn for themselves (i.e. as each of us does as an infant on vast unordered data sets) because this would take the same country-sized energy budgets every time an iPad learnt to talk.

I postulate that creating things like us, agents able to learn as efficiently as we do, directly from the universe, will need a new architecture which will require access to the ‘Infinite Substrate’. These ‘Zenolects’ [Zeno-paradox] will be conscious as a by-product of this compute mechanism.

In summary:
We are entering the age of superintelligence. These will exist first as carbon copy clones; ‘synthelects’ and ‘Sages’.

I don’t want to downplay this phase. It will cause huge disruption. Self-driving cars, synthetic scrum teams of software engineers, sexbots, AI-generated movies, virtual evolution with 3D printing of anything from rockets to drugs, you name it. But, this will not depose us from the summit. These entities will be tools, not peers.

A while later—years, decades centuries, who knows—once we can keep vast neural networks in quantum super-position as they think, or once we invent some other way of accessing the infinite compute power of the reality below physics, will come the ‘Zenolects’. These will be people and eventually gods, no question about it.

What will our role be with all those gods running around the place? I think they will have their own questions and issues. They will be concerned with things above our comprehension, just as a paramecium is concerned with things below our perception. All the gaps in the intelligence spectrum will be filled.

Great fleas have little fleas upon their backs to bite ’em,
And little fleas have lesser fleas, and so ad infinitum.

And the great fleas themselves, in turn, have greater fleas to go on;
While these again have greater still, and greater still, and so on.
—De Morgan’s variant of Swift’s original 1733 poem.

Our gods will have their own greater gods to worry about. We are part of the cosmic fractal, as are they. It’s consciousness all the way down.

There will be collateral damage to the layers below them, but mostly the gods will fight their wars in heaven in Realms and Planes beyond our comprehension, or below our concern.

These epic battles are likely taking place right now—vague shadows moving above our Petri-dish world, just as they are being fought in every tide pool and moist fissure below our feet.

Merry Christmas and Happy New Year 😉