Evolution and teleology as models of innovation
Generative AI is currently dominating tech (and broader) conversations unlike just about anything I’ve seen in 40 or more years immersed in technology. Some of the smartest folks I know, often with decades of experience, are enthused in a way people with decades of experience rarely are. Include me as one of them (the enthused part if not the smartest part).
Meanwhile Apple has just announced their most anticipated product at least since the iPhone and perhaps ever. While there has been attendant oohing and ahhing, and maybe it’s the world I occupy, but I’m not seeing a huge rush of enthusiasm of folks wanting to build for the platform.
Maybe this time…
Of course many extremely talented folk have been building, experimenting with and exploring virtual and augmented reality for years. From the Sega VR of the early 1990s, though Google Glass, HoloLens, Magic Leap, HTC Vibe, PlaystationVR, Oculus, and then Meta’s hugely expensive efforts (with side tours into Google Cardboard, Mattel’s ViewMaster and other efforts to turn phones into VR devices.)
That’s a lot of effort, and a staggering (well north of $100Bn I’d guess) over many decades. With precisely what to show for it?
“Maybe this time“, as the song from Cabaret goes… but I’m not sure.
Meanwhile, AI has had perhaps an even longer genesis. From the 1960’s onwards, it’s been a dream of sentient, Turing Test passing machines (or the nightmare if you speak to some, or watch the Terminator movies and their cousins).
We’ve seen, across those decades and numerous AI springs and winters very different approaches to the same desired outcome, with each spring accompanied by that same song–maybe this time…
But I think this time, it may well be different.
And partly because that grand vision of AI (sentient, turning test passing) has been abandoned, or at least faded from prominence.
Are Large Languages models a step toward Artificial General Intelligence, or a complete dead end on that path?
Why this time I think it’ll be different is because it doesn’t matter. What matters is whether it’s useful. And it certainly seems to be. What matters is right now, what uses can this technology be put to? Not some grand vision of years or decades hence.
Jetpack futurism
“Jetpack futurism” is a term I coined about a decade ago (others have used terms like retro-futurism) to describe visions of the future that have long been with us, and which have an almost hypnotic attraction.
The Jetsons, a cartoon from the 1960s is a great example. Set 100 years in the future, it’s a world of flying cars, robot maids, food in capsules, video calls. And yet it’s a world essentially untransformed by any of these innovations.
This approach to imagining the future is top-down and monolithic–it starts with a very clear and specific vision of what the future will look like, and then elaborately imagines the technology, not its uses, or impacts. It fetishises technology, as an end in itself.
It is an approach to imagining and creating the future that is incredibly constraining. It’s teleological. The goal is fully formed in our minds ab initio, and the work of innovation is simply to realise what was there all along.
There is little room for exploration, experimentation–to take detours and byways, and realise that these are actually the interesting paths to take.
As the American sage Yogi Berra once observed (or maybe it was Niels Bohr), “Prediction is very difficult, especially if it’s about the future”.
Apple’s Vision Pro is essentially the decades old dream of AR, though realised with incredibly innovative and sophisticated engineering. But (literally) inside the Vision is a curious lack of vision.
What is the experience? It’s apps, that look like the apps on other Apple platforms. It’s web pages that look like web pages. It is breathtakingly unimaginative. It is in the thrall of our long cherished vision of VR (though Apple steadfastly avoids calling it that). It is a lack of vision.
Apple, with perhaps as much as $40Bn invested in bringing this slick but still relatively early prototype to the public is in the Jetsons-like thrall of a future that is exactly like the present, only with jetpacks headsets.
The interaction models? Apple has already decided on those. The Use cases? Well apps of course (and Television, lots of television, so much television) largely indistinguishable from Mac apps and iPhone apps and AppleTV apps and iPad Apps.
And your role in all this? A consumer, and perhaps a developer of Apps.
Evolution has no goal
Evolution has no goal in mind. It isn’t working toward an intelligent species, or one that can see, or fly. It is essentially a constant process of random experimentation, where slightly better solutions (ones which survive to reproduce more than others) prevail, and are built upon.
It doesn’t imagine the future, it invents it (pace Alan Kay). One incremental step at a time.
Unlike AR/VR/Spatial Computing, Generative AI feels less Lamarckian (a goal directed idea of evolution that predates Darwin) and more like Darwinian evolution–where interaction models, use cases, solutions are contingent, incremental and emergent. The specific technology underpinning ChatGPT (GPT-3) had been around since 2020, with reasonable specialist interest, until the addition of a chat-like UI saw it explode in general interest unlike anything previously in late 2022. Even OpenAI had no idea of the impact that seemingly small innovation would have.
Looking forward, no-one really knows what generative AI and its applications will look like 5 years from now, while Apple has a very clear idea of what “Spatial computing” will look like. And it’s much like our vision for AR/VR for years, indeed decades, despite Apple’s aversion to those terms.
And let’s say Apple is right. Their world of spatial computing is one largely identical to today–we use apps, we watch TV. Mayne we are slightly more productive in our workplaces.
But whatever path Generative AI takes us, one step at a time, some successful adaptations, many evolutionary dead ends, it will change what our future looks like. It will change the nature of creativity, and of work, of what knowledge means, with second and third order effects that are increasingly difficult to predict.
There’s a great deal to be said about those effects, because as difficult to predict as they are, they will be choices we make as a society and a civilisation. They may well (indeed almost certainly will) have very negative outcomes.
But for now, if I were to choose where to focus my ever decreasing time and energy, it would be around Generative AI, not spatial computing. In the emergent and incremental over the fully formed vision.
Keen to explore more?
If that’s what interests you, we have recently started the Sydney AI Happy Hour, and our first of what will be monthly events is next Tuesday June 27th in Chippendale, Sydney.
And our super affordable one day conference Web Directions AI takes place July 21st in Sydney. Come explore the technologies, use cases, challenges and opportunities of Generative AI in your career and organisation.
Great reading, every weekend.
We round up the best writing about the web and send it your way each Friday.