Does AI want to be humane?

01 Mar Does AI want to be humane?

Published on Medium at https://medium.com/@nbloom/does-ai-want-to-be-humane-1ea354e6832e#.gqhttg99i

@NikolasFuturist except the fear he’s conjured in the public about AI is about AI becoming humane. #BCTECHSummit

— Noah Bloom (@nbloom) January 19, 2016

I had this little exchange recently around a Vancouver tech conference. Here are a few reasons that there’s a common fear about AI: sci-fi movies, the Turing Test, and Kurzweil.

Is it ever easy to overanalyze these sci-fi movies about their critique of our culture or our future:

Why does Alex Garland’s 2015 sci-fi flick Ex Machina show that the pinnacle of AI is convincing human-like emotion? Or that the flaw of the human is their susceptibility to their emotions or to their empathy? And, did that Ava character pass the Turing Test? Well, in a condensed, feature-length film tackling big issues and trying to entertain, ambiguity is your friend.

Feelings! Ava from Ex Machina

“The Turing test is a test, developed by Alan Turing in 1950, of a machine’s ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human.” The Turing Test is often seen as the definitive threshold for AI to be “true intelligence” and gets frequently name-dropped throughout pop culture.

Programs are already beating and tricking humans at chess and games, music composition, and even poetry. Disappointingly, often, AI programs will pose as a foreign-speaking child to slacken the judge’s scoring. Sure, it’s likely this confused and unintelligible “person” just doesn’t know very much!

I’d prefer to hear from voices in AI like Stuart Russell, Marvin Minsky, and Nick Bostrom who would argue that the Test is essentially worthless and is a distraction from the real work of AI. The only people actually working on passing the Turing Test are doing so as a hobby (eg. Ex Machina).

Minsky called one Turing competition “obnoxious and stupid” and even put up his own prize to shut it down.

Passing the Test was not meant as the goal of AI, but as a thought experiment to incite people who were skeptical of intelligent machines, or to prove that they could be intelligent through their behaviour and being indistinguishable from humans, not by being self-aware.

The players in our AI future, or rather machine learning, program-as-hardware, are likely to be the big guns: Google, Amazon (Lab126), Microsoft, IBM, and maybe even Andy Rubin’s new venture Playground. These are not hobbyists. There are different motivations. They are here to create value from machine learning for industry and for consumers. I’m particularly excited for the niche, consumer, mainstream market, every-single-day AI bots, like the expanding Slack bots (you spend your whole day in there anyways), Messenger (slowly usurping all your social chatting), and Fin (new contender from Sam Lessin and Andrew Kortina). These are not just cocktail party tricks.