AI gets real (sort of) in the enterprise

Two years ago artificial intelligence (AI) reached the peak of absurd expectations, as I tried to capture in this post. Today? Well, reality seems to have crept in, something that shows in how companies have been approaching AI, which has been to focus on low-hanging fruit, rather than moonshots.

This is according to Deloitte’s State of AI in the Enterprise, 2nd Edition, which unveils a world increasingly serious about AI. That said, there are still some head-scratchers in the data. Let’s dive into the report.

Getting the boring stuff right

First, of the 1,100 executives surveyed by Deloitte, a whopping 82 percent feel that they are getting a positive financial return from their AI investments. Of course, not everyone has benefited equally. The only industries claiming to get a high return from their investments are Tech/Media/Entertainment/Telecom, Professional Services, and Industrial Products (this last category spending the least yet still gaining much). Life Sciences and Health Care, Government, Financial Services, and Consumer Products, by contrast, all report lower ROI. Yet apparently “lower” ROI doesn’t mean “nonexistent.” Pretty much everyone who invests in AI claims to be happy about it.

The question is why.

No, it’s not because enterprises are finally able to shed their least productive employees. Executives listing “Reduce headcount through automation” increased just two percentage points between Deloitte’s late 2017 and late 2018 surveys (from 22 percent to 24 percent). Meanwhile, the far less controversial (and arguably more useful) benefit of “optimiz[ing] internal operations” climbed six percentage points (from 36 percent to 42 percent).

Much more worrisome, however, was that AI didn’t seem to improve product development and, by extension, customer experience. While 44 percent of those surveyed picked “Enhance current products” as a benefit of their AI investments, that was down from 51 percent in 2017. What about “Create new products”? Also down, from 32 percent to 27 percent. Indeed, more than half of the cited benefits dropped from 2017 to 2018, and the only other benefit to register at least a two-percent improvement (beyond those listed above) is “Capture and apply scarce knowledge.”

The positive spin on this data is that the downward trend in near-term expectations may augur well for AI’s longer-term impacts. In addition, as Deloitte reckons, the focus away from improved products may indicate a needed dash of realism:

This shift toward internal operations has been accompanied by a somewhat reduced emphasis on integrating AI into existing products and services, although that remains the most popular objective. In fact, operational change is often required before such integration can take place. Our respondents may be realizing that they should make operational changes first.

This is spot on. AI isn’t magic pixie dust that enterprises sprinkle over legacy processes and legacy tech to make them “modern.” Rather, it’s a fundamental rethinking of how to do business. Unless a company is willing to change its culture (and associated processes), no amount of AI investment is going to work.

The numbers don’t add up

Despite this welcome realism, not all of the data makes sense in the report. For example, when asked to respond to this prompt—“Relative to competitors, respondents say their company’s adoption of AI has allowed them to …”—one might expect a spectrum of responses, but somewhere on that spectrum should be “We’re not beating our competitors because of AI.” Instead, the results were as follows:

• 16 percent - Catch up
• 20 percent - Stay on par
• 27 percent - Edge slightly ahead
• 28 percent - Widen a lead
• 9 percent - Leapfrog ahead

So, no one is falling behind? Apparently, we are all above average when it comes to AI. Sure, maybe the 1,100 people surveyed all represented completely different industries and everyone not surveyed is about to be overrun by these AI-driven titans of industry but... maybe not? Maybe, just maybe, that 16 percent haven’t, in actuality, caught up. Maybe that 20 percent have fallen behind. And so on.

Perhaps it’s these rose-colored glasses through which executives are looking that accounts for the next surprise: 88 percent of those surveyed intend to increase AI investments in the next year (meaning 2019), and a full 54 percent expect to increase it by more than 10 percent. Remember how low ROI was for at least half of the industries? Well, that hasn’t dampened enthusiasm for spending more, apparently. Not to worry, though! As Deloitte writes, “Less than 50 percent of surveyed companies measure key performance indicators necessary for gauging financial returns accurately.”

In other words, “We have no idea if our AI investments have been productive but hopefully they have so let’s pour more fuel on that fire.” It would be easy to take this uncharitable view but for the increased pragmatism in the expected benefits. So long as organizations are focusing on incremental process improvements, and not on expensive moonshots, we should eventually see organizations capable of making those moonshot investments, thanks to more prudent investments today.