Everybody seems to want to get in on the machine learning hype these days. According to the head of Google’s DeepMind team, however, “There’s only a few hundred people in the world that can do that really well.” No offense, but you’re probably not one of them. No wonder that Gartner has machine learning at the absolute apex of its 2016 Hype Cycle. Nothing else promises to do so much to transform humanity…with such an anemic record to show for itself.
This isn’t to suggest that artificial intelligence has a bleak future, but rather that we’re getting way ahead of ourselves in terms of what it’s delivering today.
The magical panacea
Maybe this wouldn’t matter much, but over-inflated expectations can lead to crash-and-burn on investments when reality starts to bite. Ampify Partners’ David Beyer described what’s at stake:
Too many businesses now are pitching AI almost as though it’s batteries included. I think that’s dangerous because it’s going to potentially lead to over-investment in things that overpromise. Then when they under-deliver, it has a deflationary effect on people’s attitudes toward the space.
We’re seeing this across the market, generally. Developers are already starting to grow wary of chatbot claims, as one example. Chatbots seem to resonate in China and throughout the Asia-Pacific region, but their application to Western markets has been underwhelming, in large part because, well, they’re sort of dumb.
They also aren’t likely to get better in the near future, as Alec Pestov has suggested: “[A]fter hundreds of millions in R&D, Cortana, Google Now, and Siri are still quite inadequate at understanding natural speech. It will take years before machines are capable of understanding human speech to the degree necessary to correctly process the nuances of conversations.”
And yet, we hype them, leading 25% of developers to view them as a waste of time, according to a comprehensive developer survey. They’re not a waste of time. They’re simply not ready for the burdensome hype we’ve placed upon them, and on artificial intelligence and machine learning, more generally.
It’s just math, folks
Ironically, Beyer says, much of what we attempt to do today with newfangled machine learning could actually be better served with simpler approaches: “The dirty secret of machine learning…is so many problems could be solved by just applying simple regression analysis,” or, as investor John Ryu added, “a handful of if/then statements.”
Basecamp data scientist Noah Lorang argued much the same thing: “The dirty little secret of the ongoing ‘data science’ boom is that most of what people talk about as being data science isn’t what businesses actually need….There is a very small subset of business problems that are best solved by machine learning; most of them just need good data and an understanding of what it means that is best gained using simple methods.”
Those simple methods? “SQL queries to get data, … basic arithmetic on that data (computing differences, percentiles, etc.), graph[ing] the results, and [writing] paragraphs of explanation or recommendation,” Lorang said.
In other words, we’re introducing unnecessary complexity to systems that we can barely make use of even in their more primitive states. As much as we may want to go “all in” on machine learning, most companies haven’t even figured out how to de-silo and clean their data (or do basic analysis of that data once the preparatory work is complete). Without that initial step, no amount of machine learning software will magically transform a business. Instead, it will cause enterprises to question the machine learning investments in the first place, rather than tackling their approach to machine learning, which is really the core problem.