What Google executive Schmidt gets right, wrong about future of artificial intelligence
Over the weekend, Alphabet executive chairman Eric Schmidt (you probably know him better as Google executive chairman Eric Schmidt) wrote an op-ed on artificial intelligence for the BBC. While he does a good job of describing ways artificial intelligence will drive forward some key consumer segments (specifically, music and travel), he doesn't go far enough in developing some truly breakthrough ideas about where AI's headed.
That's especially surprising, given Google's former emphasis on "moonshots" and longtime reputation as an innovation leader.
Take, for example, Schmidt's remarks about "smarter music," which some people in the tech community have perceived as a shot across the bow at Apple Music. Schmidt says that intelligent algorithms can do a better job than human tastemakers at picking music: "A decade ago, to launch a digital music service, you probably would have enlisted a handful of elite tastemakers to pick the hottest new music."
"Today," he says, "you're much better off building a smart system that can learn from the real world -- what actual listeners are most likely to like next -- and help you predict who and where the next Adele might be."
Schmidt also says that AI has a real role to play in sorting out "real-world messiness." The example he gives is selecting a last-minute vacation that will satisfy everyone, even "two picky kids," while still fitting within the constraints of a family budget. If you add enough data about preferences and budget parameters, an AI algorithm could optimize the perfect vacation choice. That's interesting, but it also sounds like something that's possible today _ not a reach goal for the future.
And it's much the same for other consumer segments. Schmidt says that AI means "smarter filters on your emails, your social media feeds, your schedule." It means more ways to apply AI to everyday life -- such as the automatic captioning of family photos -- and an easier way to manage the flow of information around us.
Good points, but Schmidt also cherry-picks areas where Alphabet is poised to have the most impact and areas where Google can start putting its AI acquisitions to work within the fastest possible time frame. In 2014, for example, Google acquired the much-vaunted AI startup DeepMind at a reported price tag of $400 million.
You can almost see Schmidt's mental gears whirring, as he thinks about each new company within the Alphabet parent company, and how AI might be applied to each of Google's businesses in order to appeal to stakeholders. Music, check. Search, check. Email, check. Rather than leading to something completely transformative -- think intelligent robots or self-driving cars -- AI may end up being more of a special sauce that's sprinkled over the company's core assets.
And that's a shame if AI is just about optimizing Internet ROI. Especially since Schmidt's earlier book, The New Digital Age, which he co-authored with Google's Jared Cohen, gives so many examples of how Schmidt is truly taking a big-picture view of the digital world. In that book, Schmidt describes how the Internet is changing the future of identity, citizenship, government, journalism, revolution, terrorism, war and foreign policy.
It's likely that AI, if it continues to progress at current pace, will have just as an important impact on the future -- and not just on consumer-facing segments of the Internet. If anything, Schmidt aims too low in his BBC op-ed by focusing only on the low-hanging fruit, such as email.
Where Schmidt misses the full potential of AI are three areas that he specifically cites within the BBC op-ed: genomics, climate change and energy. Remember when we heard about Google's new renewable energy initiatives, bold attempts to change health care and wacky Google X projects? That seems to have all been replaced by a much more pedestrian concern about making sense of all the data, turning AI into just a powerful data-crunching machine.
Moreover, Schmidt mentions AI legend Geoff Hinton _ one of the people that he cites as an inspiration for Google's current AI initiatives _ but doesn't mention the legendary Ray Kurzweil, who was recruited by Google to work on AI projects. He alludes to Google's Gmail, but doesn't mention DeepMind, even though DeepMind is now teaching machines how to read.
Are those intentional or unintentional oversights? Sometimes the biggest clues are from what's not mentioned, not what's mentioned.
If the big news story about Schmidt's op-ed on AI is the turf war between Google and Apple Music -- as Mashable and re/code suggest -- it seems that something is being missed about the full potential of artificial intelligence. If AI is just about making consumer lives less messy, then there's maybe not as much to get excited as was once thought.
Or, this op-ed mentioning everyday consumer needs -- photo captions and spam filters -- could just be an attempt by Schmidt to downplay all the dystopian fears about the future, while still playing up the long-term potential of artificial intelligence. At SXSW earlier this year, for example, Schmidt suggested that many of the dystopian views of the future are being overhyped. Adding captions to photos, how harmless could that be? (Wait, wrong question to ask, unless you like demon puppies.). Or better spam filters, who's opposed to that?
This view of AI, in which artificial intelligence is just another digital innovation that's going to "reduce the noise of everyday life" raises the inevitable question: What if the leading minds within the world of artificial intelligence promised us superintelligence, but ended up just giving us better spam filters?