AI Threat Is Tech's Fart in the Room

I was at IoT World in Dublin this week and something Google said at the Informa-organized conference left me feeling distinctly queasy.

It was during a breakout session on analytics, and Dvir Yuval, who heads up EMEA online partnerships for Google Cloud, was talking about AlphaGo. For anyone who's been living in a cave this year (and with Trump and Brexit to worry about, that's probably advisable), AlphaGo is the artificial intelligence Google (Nasdaq: GOOG) developed to challenge Lee Sedol at Go, a board game that seems to have originated in ancient China. Go is fiendishly complex, by all accounts, and Lee Sedol is one of the world's best players. (See Google: Little Chance Today's Startups Become Tomorrow's Giants.)

AlphaGo versus Alpha Male
Google's AlphaGo AI famously beat Lee Sedol, one of the world's best Go players, during a contest in March this year. (Source: flickr/Buster Benson)
Google's AlphaGo AI famously beat Lee Sedol, one of the world's best Go players, during a contest in March this year. (Source: flickr/Buster Benson)

Disappointingly, for anyone outside Google, AlphaGo won 4-1. The victory was particularly troubling for supporters of homo sapiens because Go, according to experts, demands far more creativity and intuition than chess. What Yuval finds especially intriguing is that some of AlphaGo's moves made zero sense to commentators and other world-class players. In soccer terms, they were like a player "kicking the ball out of the stadium," he says. Except they weren't, because they ultimately allowed AlphaGo to win. As far as Yuval is concerned, this suggests that AI could eventually push the boundaries of creativity in a way humans cannot.

Now this may be something Google and those in the AI community applaud, but I'm inclined to view it in the way your typical triceratops might have looked at the approaching asteroid. Machine-generated creativity and ingenuity is a kind of extinction event. It might not wipe us out (initially), but it will guarantee our species' uselessness. Imagination and innovation are the last refuge of mankind. Once robots become more innovative, we are pretty much finished.

This doomsday scenario will be familiar to science fiction fans everywhere. What's much scarier is the obvious pace of AI development in science fact. Five of the world's ten most valuable companies are now technology firms heavily focused on AI and run largely by machines. Facebook, one of them, is currently worth $347 billion on the Nasdaq and employed fewer than 12,700 humans at the end of last year. To put that in context, carmaker Ford, which had nearly 200,000 humans on its books in 2015, has a value of about $48 billion today.

Want to know more about the Internet of Things? Check out our dedicated IoT content channel here on Light Reading.

Ford is clearly overstaffed, but even Facebook relies too heavily on humans. In future, a small number of senior executives, dealmakers and science geeks will preside over a much more automated business. Computers will use big data to decide which new features should be introduced, and then design algorithms to support them. Many of today's employees will find their only interaction with Facebook is as consumers.

Few are prepared to face up to this, however. To borrow a quote from Canadian politician Michelle Rempel, the issue has become like "a fart in the room" -- something no one really wants to discuss (Rempel was referring to the Canadian province of Alberta when she gave vent to the simile this week). Sure, we've had plenty of dystopian visions from authors and filmmakers, as this excellent blog from Light Reading's Brian Santo points out. But the industry giants are either pretending that nothing smells rotten, or simply don't care. (See How to Speak Knowledgeably About the AI Threat.)

Is Google Skynet?
Terminator 2's Miles Bennett Dyson (pictured, as played by Joe Morton in the movie) develops an AI called Skynet that starts a war against humanity.
Terminator 2's Miles Bennett Dyson (pictured, as played by Joe Morton in the movie) develops an AI called Skynet that starts a war against humanity.

Either explanation is conceivable. For the next few years, at least, Google and our other technology overlords will be the big winners from machine learning and AI. That's why Yuval can get excited about AlphaGo's creativity and not fret about its implications.

Like the growing use of fossil fuels, AI's rise seems inevitable -- something global powerbrokers are pushing for even though it will eventually kill us. Google is the oil tycoon who never talks about climate change. Sooner or later, though, it will have to confront the bogeyman. In this year's populist upheavals are signs of frustration at an uncaring establishment and the yawning wealth gap between an uber-rich minority and the poor masses. Globalization, free trade and immigration have been cast as the villains by the likes of Donald Trump and the UK's Nigel Farage. Yet machine learning and the automation it brings will fuel the popular anger, hurling more on to the scrapheap of humanity. AlphaGo shows we are all at risk.

— Iain Morris, Circle me on Google+ Follow me on TwitterVisit my LinkedIn profile, News Editor, Light Reading

COMMENTS Add Comment
CTO84020 12/6/2016 | 7:32:13 AM
Re: Facebook relies too heavily on humans I wonder if the ad revenue lost by quashing fake news is affecting FB's resolve to get this done. This is a problem that could and should have already been solved.  Be it AI, algorithms or brute force human intervention, this is not a complicated problem, but a voluminous one.
Joe Stanganelli 11/29/2016 | 7:23:18 PM
Re: Bridge @Michelle: I have supposedly one of the best bridge computer games, and I find it woefully inadequate, inflexible, and substandard.

BBO (BridgeBase Online)'s robots aren't bad, but the problems aren't with play so much as they are with bidding -- which requires MUCH more "intelligence" than the basic logic of simple card play involves.
Michelle 11/28/2016 | 10:59:54 PM
Re: Bridge Really!? I thought more progress had been made by now. What game will AI tackle next?
Joe Stanganelli 11/28/2016 | 8:14:51 PM
Re: Facebook relies too heavily on humans @James: And then there's the whole subjectivity/poltiically motivated problems of defining what "fake news" actually is.
Joe Stanganelli 11/28/2016 | 8:12:40 PM
Re: What AI? @t.b: Indeed, logic is logic -- and computers are made for logic.

I'll be more impressed if a bridge-playing computer can solve the ACBL's monthly bridge-bidding problems.
Joe Stanganelli 11/28/2016 | 8:10:44 PM
Bridge For what it's worth, programmers have yet to invent a similarly strong bridge-playing computer; they all have their problems and are far from perfect.
t.bogataj 11/28/2016 | 3:25:18 AM
What AI? Let's name things properly. This is not AI, just as chess-playing programs are not. They're algorithms, not intelligence.

And please, do not use the word "creativity" in the context of algorithms.

Michelle 11/26/2016 | 8:34:48 PM
AI for the win This is a really interesting test of AI. I too hoped the human would beat the machine. 

*Who wrote this article title? Clearly, LR editors have a lot of fun behind the scenes :)
James_B_Crawshaw 11/24/2016 | 12:34:36 PM
Facebook relies too heavily on humans Facebook's recent problems with fake news demonstrate the problem the company has with anything complex that is not carried out by humans.

Facebook's remedies to address fake news include:
1) Use third party organisations [STAFFED BY HUMANS] as fact checkers
2) Verifying information with journalists [HUMANS].
3) Relying on the community [MOSTLY HUMANS] to report fake news.

Hat tip www.radiofreemobile.com
Sign In