Very important artificial intelligence research was presented last fall at the Oaklins Desilva+Phillips 2017 Dealmakers AI Summit, an elite invitation-only event in Manhattan to which a friend wrangled plebeian me an invitation.
The research, "Reshaping Business with Artificial Intelligence," produced by MIT Sloan Management Review and The Boston Consulting Group (BCG), reveals stark differences between companies that are leading the charge to adopt AI ("Pioneers") and those that are not ("Passives").
From my own point of view, the report lends credence and a strong fact base to a favorite theme I've pounded on in previous columns: Companies that don't begin working with AI now will never be able to catch up to their competitors who do.
Most of the 3,000+ respondents across 112 countries and 21 industries, of course, are NOT Pioneers. Only 5% have "extensively incorporated [AI] in processes and offerings," while another 18% have incorporated AI into "some" processes and offerings. That makes for a total adoption level of 23%. Another 23% are only piloting some AI projects, which leaves 54% who have not yet taken even that step.
Based on adoption level and questions that get at how well the respondent understands AI, 19% of the study respondents ended up in the Pioneers pool -- they both understand AI well and are adopting it. Next, at 32%, are Investigators, who understand AI but are still piloting; followed by Experimenters (13%), who are "learning by doing" — that is, they’re relatively clueless about AI but are nonetheless piloting or adopting in order to learn. Finally, 36% were deemed Passives -- "organizations with no adoption or much understanding of AI."
According to Martin Reeves, the senior partner and managing director of BCG who presented the findings, it is extraordinary for such an embryonic market that EVERYONE -- from Pioneers to Passives -- has very high expectations for how AI will impact their industry. For example, 84% of respondents agreed with the statement that "AI will allow us to obtain or sustain a competitive advantage."
Quips Reeves: "There's going to be a lot of disappointment and disillusionment. Clearly, 84% of companies are not going to come out of this competitively advantaged."
Importantly, marketers are among those who may be affected most -- and earliest. Respondents from what BCG designated the technology, media, and telecom industry reported the "largest effect" when asked to rate, on a five-point scale, "To what extent will the adoption of AI affect your organization's offerings today and five years from today?"
Although I don't have access to the underlying data, a close read of the charted data shows about 21% or 22% of the tech, media and telecom group reported a large effect today, climbing to about 72% or so who anticipate a large effect five years from now. That was the most of any industry, both for today and tomorrow.
Differences in understanding AI are where the biggest challenges lurk for those organizations not yet adopting the tech. As Reeves explains, Pioneers are wrestling with real-world issues, bumping up against privacy and regulatory concerns, and challenges like finding and hiring AI talent. Passives, meanwhile, are dithering, not understanding how to make a business case for AI, and facing no executive support nor tech DNA.
In other words, every day, and in every way that matters, the Pioneers are lengthening their lead over the Passives, like Secretariat in his never-equaled, epic 1973 Belmont Stakes run, still the world record for the mile-and-a-half on dirt (2:24).
Reeves emphasized a big misconception Passives have about the relative value of AI algorithms and data sets. Of note, subsequent speakers achieved consensus around the idea that 70% or more of the hard work and value creation in AI apps comes from the data you need to train the algorithms — and then the data they subsequently analyze. By comparison, algorithms are relatively cheap and often open-source.
But Passives seem to think when they finally decide to do AI, they can buy some AI software and be off to the races. They don’t understand the hard work that must go into TRAINING AI systems. You have to “raise” them, like children. Organizations that lack data experience and prowess may soon find themselves afloat on a river of excrement without an explicit means of propulsion.
Over time, notes Reeves, competitive advantage will accrue to those organizations that own high-quality data sets, that develop the ability to learn faster than their competitors, and who "experience effects."
There are more worthwhile insights in the study, which is fairly broad and far-ranging, and I hope to unpack more in future columns.
Meanwhile, remember: Just do AI. It's good for you.