Recent internal IBM documents suggest that the company’s Watson product made “often inaccurate” suggestions, raising “serious questions about the process for building content and the underlying technology.” One doctor was more blunt: “This product is a piece of s—-.” For those quick to laugh at IBM’s troubles, the reality is, as Blair Reeves points out, “Given the pace of deployment of existing technology to industry…’AI’ is [unlikely] to instantly transform much for the foreseeable future.”
In other words, we’re nowhere near the AI future so many celebrate (or fear).
Buying more fear
Yes, fear. You’d think we’d be getting comfortable with artificial intelligence (AI), given how awash the media is with mentions of machine learning (ML) and AI. Nope. As Gartner analyst Craig Roth has highlighted, we’ve seen “AI headline hyperbole” dramatically expand in 2018: “Destroy” is up 163%, while “Kill” has swelled by 58%.
SEE: IT leader’s guide to the future of artificial intelligence (Tech Pro Research)
Yes and no. As much as we may like to play up a scary future of robotic overlords, companies keep investing in AI, desperate to find some competitive advantage in the next algorithm. Indeed, as KPMG has calculated, we can expect $232 billion in AI investments by 2025. Apparently we’re a bit queasy about this AI future, but particularly with the thought that our competitors might reach that future first.
Given the current state of AI, however, maybe the fear is misplaced.
The future is slow
For every success story you read about some company transforming itself through AI, there are 1,000 more stories of companies either not ready for AI or failing to get AI efforts off the ground. The irony is that many of these efforts are unnecessary. As Basecamp data scientist Noah Lorang has argued:
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…[like] 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.
Math isn’t sexy, however, and so companies continue to strain at AI. Fortunately, as O’Reilly’s Ben Lorica has noted, while most companies are still in the early stages of kicking the tires on AI, needing to “wade through an array of methods and technologies, many of which are still very much on the leading edge,” there is good news in that “[S]ome companies at the forefront are beginning to share best practices, tools, and lessons learned as they deploy AI technologies.”
In other words, there’s help out there, if you know where to look for it.
Some will immediately jump to the conclusion that the best source of AI wisdom will come from companies like Google that are the most advanced (and, fortuitously, that are building a cloud business on the premise of giving access to its data science smarts to mere mortals). The problem with this approach is that Google is way ahead of most companies. The other problem is that Google may not be using the “sophisticated” AI enterprises are expecting. As Hassan Khan has noted, “I’ve heard that up until just a couple years ago Google was using logistic regression for its search results.”
True or not, it’s smart for most companies to do as Lorang has suggested and take a simple approach to AI. The sci-fi AI we celebrate (or fear) in the media is so far off into the future that companies can waste lots of time and resources chasing it. Far better to get “good data and an understanding of what it means that is best gained using simple methods,” as Lorang pointed out.
Is your organization planning to launch any AI projects in the near future? Share your experiences and advice with fellow TechRepublic members.