How Google is Remaking Itself as a “Machine Learning First” Company
Steven Levy
2.3K61

Google turns wrong. In the 1950’s during the Cold War, there was a program to win a nuclear war fought on US soil. By the 60’s they’re out of power, Minsky and thus Turing (his work) become more important to JFK with MAD. But in the 50’s as part of that program, there is a plan to have self-driving tanks. Somehow, all the other crazy shit from LSD-25 cigars to hypnotizing prostitutes fades away, but this plan, which included having 1/3 of all DoD vehicles autonomous by 2015, stays on the books.

Google bought into the DoD and Darpa in a big way. Who knows, maybe they got favors in return? Maybe strategic bridge loans? Maybe they had top-secret gov’t contracts at a time, like during the dot com crash, that allowed them to survive and then thrive? Maybe we’d all have HotBot phones instead? Ask Jeeves?

AI, computer-vision, parallel processing, multi-thread processing, Turing-machines and thus virtual machines and thus Java or Unix and bots, chatbots, all related. But then that predicates IoT, which predicates big data, analytics, which predicates less-powerful processors, which runs counter to Moore’s Law.

I do not believe in Moore’s Law as such, but rather than a Law more like a rule, and one which involved a certain amount of groupThink collusion between Wall Street and the Valley. I always predicted it breaking down by the time it hit the nano level, leading to: carbon based computing, AI, nano-tube medical tech implants, further attempts at cyborgs, monkey testing, DNA experimentation leading to… well, Brave New World. Thus legal restrictions with respect to trying to create a Master race with eugenics.

But the point with Moore’s Law is that everyone agrees the hidden part of it is the Market will always want faster cheaper more processing power. Intel is built upon investing in new foundries. IoT creating demand for 'stupid' chips blows the paradigm.

We’ve been in a wave involving Intel layoffs, desktop slowdown, rise of IoT, Stanford pipeline, Andreseen angel investor network that should have led to smart homes, smart TV, AR, which sure, are there, but why did nest fail, why hasn’t Apple TV already crushed it, how come HoloLens is always a demo and for next year? Maybe I’m wrong, but in the 80’s it was all about the CPU, then it was price point delivery for IBM clones, then GPU’s and peripherals etc etc the browser wars, search and so on.

If IoT which is predicated on AI doesn’t monetize, or have a break-out and soon, pacman, pole position, the walkman, the iPod then all that VC might dry up, and if there is a 2008 style market crash? What happens to Intel? If Google’s bet on AI is related to self-driving cars and if that is related to a 50’s crazy DoD plan to win a nuclear war with the Soviets on US soil via some long forgotten top-secret E10 plan predating darpa, the arpanet and internet, maybe they have no choice, made a deal with the devil?

They’ve yet to make a major mistake, they’re still primarily an ad company, something like 20% of their operating income still comes from the Google homepage plain-vanilla paid first sponsored ad link.

But if they bet wrong, a lot of people follow them. If tensor chips don’t create their own Moore’s Law where 'intelligence' for servers and thus bots doubles every six quarters, how big a bet is it, on foundries? On the intellectual capital of r&d? Is that a limited resource? Is the whole plan predicted on a self-fullfilling prophecy where the tensor chips eventually design the next generation?

What if it all turns out to be synergy and time warner aol? Google will go bankrupt, be a great brand name, but droid will survive, as will Samsung and the Chinese factories of Foxcon.

Self-driving cars will end up in a heap with flying cars and contragravity and interstellar flight, moon bases, safe portable nuclear power, ray guns, the industry is moving on: video news, retro flip phones, porn VR, adjustment to 4th amendment privacy concerns, a debate over fair compensation for artists. Though to be fair, not all of these are mutually exclusive with the cloud, encryption, and voice search and thus IoT, bots AI and Google’s approach. To assume that even if self-driving cars don’t replace cars as the most popular form of transportation that Google doesn’t have a backup plan or can’t use and monetize vertical product chain design which would be developed in concert, the tensor chips or even the social engineering they’re currently creating to 'build' these new ninja designer e.e.’s would be as foolish as betting on Amazon’s 1999 acquisition petsdotcom over its other purchase from the same year, Alexa.