Interrogating the Hype Cycle

“Imagine a place and you will be able to step into it. Conjure up a dream and you’ll be able to fly through it… It’s a computer-generated world where you see and move and feel… Will real life ever be the same?”

- Virtual Reality, ABC Primetime April 9, 1991

With the sudden influx of funding from government and industry, the AI community now finds itself with more tools, people and energy to address the complex issues of natural language, vision and expert systems.

Mini-Micro Systems, December, 1983

AI has grown from an esoteric part-time pursuit of a few visionaries to a full-fledged science… and in a few years, the computer as we know it is likely to be dramatically transformed.

Byte Magazine v6 issue 9, September 1981

“If you believe the hype, virtual reality is virtually at your service already.”

Popular Science June 1993

So much hype. So many breathless proclamations, compelling demos, university stuff, sure, but also military and corporates, a litany of movies and books about the onrushing tomorrow where we step into our 80's neon imaginations and party with Max Headroom.

But virtual reality and artificial intelligence slumped back into the shadows and stayed there for decades.

Tech doesn’t become transformational just because everyone’s talking about it. Nor is it bound to disappear if we dismiss it. Now we live in attention economies, overwhelmed by noise and clamoring proclamations trying to draw our eyes and our wallets and our votes. The hype is amplified and distorts the picture, clouds our thinking.

The world is only noticeably changed by tech when so many people are using it that it starts to re-shape us, our organizations, cultures, and societies; when those hockey sticks of adoption become solid S-curves of broad installation.

Interrogating the Hype Cycle

It’s easy to see disruption in the rear-view mirror but to see ahead more clearly we have to look past the Hype Cycle. Situate the tech in its many ecosystems of enablers and inhibitors and the deeper needs and fears of the people who might adopt it. This is map-making for complex dynamic systems, back-casting the conditions for emergence.

AI & VR are great lenses to look at how disruption does and does not happen.

Both require tons of computation. The more, the better (a fact not lost on Amazon or Google or Microsoft). There wasn’t much until recently but now there’s massive superclusters held inside warehouses around the globe; there’s transistors chemically etched at 14 nanometers, prompting people like Autodesk CEO Carl Bass to speak somewhat breathlessly of “infinite computing.” And it turns out that GPUs with a couple of decades of twitch shooters and MMORPGs to drive their cost performance are not only great for VR (natch)but also for AI (woah). Quick to capitalize on a lucky break, GPU leader Nvidia has pushed out a whole product line — its Tesla chips — for deep learning and high-performance computing.

Data is food for AI. 1980’s-era AI was thin and hungry — weak chips, scrawny training sets. A few decades of Internet and connected things unleashed vast flows of structured and unstructured data to crawl through and classify and model. AI is anemic without it. AI is the bacterial colony growing on Big Data.

1990’s VR had no ecosystem of content tools and creators, very little usable guild knowledge. Now we‘ve got sophisticated toolchains running on local clusters with legions of 3D world builders pushing polys and making shaders. The field has matured significantly.

An affordable technological support fabric for AI & VR is now in place. Clock cycles have gone up, prices have come down, and there’s and endless river of data.

Cheap and powerful components

The global smart phone supply chain has driven intense miniaturization and performance across a broad category of components, from radio antennae to high-density touchscreens, while pushing the price of performance down dramatically. It’s a low-barrier global manufacturing platform that makes it easy to prototype, and merely a matter of money to scale and distribute. Palmer Luckey built his Oculus prototypes at home just a few years ago.

As Benedict Evans noted on behalf of venture juggernauts Andreessen Horowitz, “it’s as though someone dumped a shipping container worth of Lego on the floor and we’re working out what to make.” Indeed. Bear this in mind when evaluating drones and IoT and driverless cars.

That’s why Google Cardboard and Samsung’s Gear VR can produce a decent immersive experience — because a smartphone screen and it’s sensors are basically the same as an Oculus, minus a lot more performance. And decent doesn’t really cut it for VR. People hurl. So you have to ask, how many users will settle for the cheap seats in the VR theater? How many will dismiss it after a few lame trips on bad content?

Plenty of money

One way or another, broad adoption of a technology takes money. Actual tangible disruption only happens at scale and scale is expensive. Lots of money is flowing into AI & VR. More each year. It’s flowing out of incumbents, Big Corps, pension funds and flooding into startups and product groups and consultancies.

Money shapes the maturation of a technology and guides how it will evolve and advance. Growth cycles often start with mostly fast-moving investment capital looking for a hot place in the sun to grow. Then it shifts into slower production capital to reach scale, deliver product, and fund R&D. VR still feels likes it’s partying with early-stage exuberance and reasonable uncertainty but AI has trended significantly into production capital, maturing into product groups and case studies.

Pay attention to where the money is coming from, who’s got skin in the game, and how their strategy is (or isn’t) informed by risk. Facebook bought Oculus for $2 billion. On launch day Oculus founder Palmer Luckey tweeted 15 minutes into pre-sales that “we are experiencing insanely high load.” On their own big day, HTC said its Vive headset had sold 15,000 units in the first ten minutes. Those are impressive numbers with huge capital commitments that must be honored. Sony follows in October with PlayStation VR and a foothold on its 65 million or so PS4's installed.

Gartner predicts 1.4 million headsets will ship in 2016, and another 6.3 million next year. Market research firm TrendForce claims the VR market will reach $70 billion by 2020. Money is pouring into content. Jaunt has taken over $100 million, Wevr got $38 million to build a VR marketplace. Hollywood Old Guards like Lionsgate and Fox are getting on board the VR express.

AI is taking ridiculous amounts of capital, getting into everything with a chipset, some sensors, and network access. Google dropped $600 million for DeepMind. Google CEO Sundar Pichai says “machine learning is a core, transformative way by which we’re rethinking everything we’re doing.” Everything in Google is a lot of everything. IBM is spending billions of dollars just to feed Watson, staking its future on “cognitive solutions” by acquiring data-heavy feedstocks like The Weather Company, Truven Health Analytics, and Merge Healthcare. IBM CEO Ginni Rometty said (with something like 15 consecutive quarters of losses) that “the ability to think, learn, and understand systems, products and processes is the dawn of a new era: the cognitive era.” Microsoft, Salesforce, Amazon, Facebook, and Apple are all aggressively pursuing machine intelligence, while venture funding continues to pour into early-stage companies with a seven-fold increase from 2010 to 2015.

By the money reckoning, things are really different this time around for VR & AI. Compute and data performance, supply chain readiness, pricing, serious revenues, and massive commitments from some of the biggest companies in the world.

Talent migrations

Perhaps not surprisingly, a talent migration has followed. The wizards of AI have been fleeing R&D labs and academia and running straight to well-heeled corporate business units with insane amounts of fascinating data to play with and 7-figure AWS bills. Facebook got Yann LeCunn from NYU. Alex Smola left Carnegie Melon for Amazon. Google got Andrew Ng from Stanford, who then left for Alibaba — now a partner of Nvidia. Cambridge’s Geoffrey Hinton went to Google. DeepMind, Google’s big AI acquisition, bought a clutch of the brightest PhD’s in the field. Twitter just bought like 12 of them for $150 million. They also got Magic Pony’s IP.

With VR there’s more talent already deployed in the field at gaming studios, film and TV, and CAD companies. The migration is more like a re-pooling. Nevertheless, Apple (always a follower) hired 3D UX researcher Doug Bowman away from Virginia Tech. Given Apple’s industrial design focus, the hire suggests they still think there’s a lot of UX to work out. And they’re right.

But here’s the point: Powerful and accessible development and distribution platforms, readily available capital flows, and steady migration of talent from academia into businesses are all signals of a technology category that has the potential to reach broad adoption.

What are the barriers to adoption?

Disruption requires adoption. At the end of the day it’s about people. What are the fundamental human factors that might prevent broad adoption ?How much does it cost? Is it cheaper and better than the competition? Is it easy to install, use, and maintain? Is it fun or appropriately challenging? Does it make you look stupid or throw up?

If a technology is difficult, expensive, or uncomfortable to users, it simply will not be widely adopted. Consider 3D modeling & printing, gene therapy, and Google Glass respectively. Likewise, if it’s trivial and unnecessary (3D TV).

For VR you have to literally cover your face. Including your eyes. And often your ears. This is not a natural thing for animals to do in a competitive and selective environment. Just let all that go, find a safe place, and step into the virtual world. There aren’t any other people here but there will be, soon. If we can get the through-put and latency to play nice. By the way, Tom Wheeler would like to talk to you about the impact of VR.

AI asks us to give up control. This is also difficult for humans. Especially when it’s our jobs. Or our safety. AI is explicitly predicated on offering a substitute for humans. It can remove human operators from the decision chain. Maybe good for driving, which we’re really poor at. And maybe very bad for governance that could become seriously algorithmic and dehumanized. Look out for Terminator bureaucracy.

These things make us hesitate with VR & AI. Our new tools keep getting better and stronger, and they get here faster and faster in a world that seems to just get more and more complex.

The regulatory response

Which brings us to a large and real braking mechanism for adoption and disruption: the regulatory environment. Governance and regulation is designed to dampen change because change introduces novelty and chaos. It’s a homeostasis that helps us avoid the sometimes-catastrophic swings of complex systems. It’s what keeps children from being drafted into early labor, hobbyist drones from downing 737’s, and autonomous and assisted vehicles from running over puppies and automotive manufacturers.

The regulatory response often happens after social and economic upheaval driven by the disruptors. Uber is the poster child for the fast-moving, supply chain-enabled, highly-capitalized and user-friendly disruptor that rips through incumbent businesses, re-programs human behavior, radically re-shapes labor markets, and gets all up in the grill of regulators. It is the forlorn and broken wails of the disrupted, the incumbent tax base, that have called down the regulators, beseeching them to heel the megalithic, on-demand transportation network. Regulators must balance the demands of incumbent employers, special interest groups, the citizenry, elites, and Uber itself.

We should keep a sharp eye out for signals of regulatory response. This means things have taken off enough to have impact, and therefore the brakes will be applied to slow it down for more careful study.

No tech is an island

Situate the tech, map it against enablers and inhibitors, evaluate the stocks and flows, and wield basic human behavior to better assess its potential, its speed, its competitive threat, and its impact.

For example: Fully-autonomous cars. Barrels of money. Big companies. Smart robots aren’t that tough anymore. You can wrap multispectral sensors all over the car and run a live simulation of the road inside the car brain. Mostly easy. Convincing everyone you won’t run over puppies and children? Or emulsify entire industries on the way? Harder. Freeways might do, they’re structured paths. Cities, downtowns, pedestrians, much harder. Regulations say when, where, and how fast. People decide if robo cars are safe, reliable, cost-competitive, compelling, or an unholy affront to the Great American Mythos of Cowboy Car Culture. This is not difficult it’s just not as easy as blindly accepting or dismissing the hype.

Use these maps as a starting point to better interrogate any soon-to-be-totally-revolutionary, for-sure-gonna-change-everything tech. Wearables, consumer IoT, drone delivery, personalized artificial assistants, cloud agents, chat bots, connected cats, and every other hot-topic innovation climbing up the Hype Cycle. They’re all flashy and exciting but will we really feel their presence, their significance any time soon?

It’s actually a very long way from technological innovation to disruption. Usually the timing is just wrong. But if you watch the river closely enough, you can get a sense of how and when it will change course.

https://www.youtube.com/watch?v=rVn3H93Ysag&feature=youtu.be