When looking at enterprise AI adoption, several impediments enter the discussion. That list often includes things like organizational inertia and change management. But when it comes to AI there’s a looming adoption barrier rarely discussed: Do organizations even have the ingredients to make AI work?
By ingredients, we of course mean data. AI is only as good as the data that you feed it. And this concept of data readiness was the focus of a new study from Precisely and the Center for Applied AI and Business Analytics at Drexel University’s LeBow College of Business (say that three times fast).
One of the biggest takeaways is the gap between AI ambition and AI readiness. For example, 60 percent of the study’s enterprise survey respondents (n=565) report AI impact and interest (up 46 percent from last year). But only 12 percent report that their data is sufficient for AI implementation.
Going deeper into the rift between aspiration and readiness, 76 percent of respondents say that data-driven decision-making is a top organizational goal. And they recognize that AI is the path to that goal. However, 67 percent don’t completely trust the underlying data – up from 55 percent last year.
Trust Issues
That trust factor is critical, and low ratings signal a lack of confidence in AI inputs. Without reliable data being fed into AI systems, the outputs will always be dubious. It’s a classic “garbage in / garbage out” scenario. And it says a lot when the “garbage” sentiments come from within, as this study demonstrates.
In that light, 77 percent of respondents rated the quality of their data as average or worse. That’s up from 66 percent from last year’s study. Other concerns include lack of adequate tools for automating data quality (49 percent), inconsistent data definitions and formats (45%), and data volume (43%).
Meanwhile, a key question that emerges from all of the above is what’s the source of these data issues. 62 percent of organizations believe that some of their data challenges, and lack of confidence, stem from a lack of data governance. That includes things like how secure it is, and where it’s being stored.
Another source of doubt among organizations – to the tune of 42 percent of responses – is a shortage of the right skill sets. When playing with fire – as AI is seen to many rightly risk-averse executives – specific skills are required. And AI aptitude is relatively scarce considering its early stage and high demand.
Foursquare Elevates its Attribution Engine with Sales Impact
Where it Matters
Synthesizing all of the above, it’s good that a degree of measured doubt is in place where it matters – enterprise stakeholders. Like many other hype cycles, unsexy realities like data integrity and security don’t get the same headlines as grandiose proclamations about the tech’s world-changing impendence.
Usually, later in those cycles – after the “trough” or shakeout – these unsexy topics are given proper and sober diligence. In the meantime, the right people are considering such concerns which is a good thing. Irrational fear of new tech is one thing, but measured skepticism and risk management is another.
As for the trajectory of AI’s hype cycle, its correction won’t be as dramatic as past vapor-filled hype cycles (we’re looking at you metaverse) because it’s real and it’s here today. But it’s still probably over-subscribed. And though no one knows the timing, a correction is probably coming at some point.
Meanwhile, market developments will mitigate some of the above enterprise concerns. For example, talent shortages will be alleviated as demand-driven skills pipelines attract new technical agility. And when AI’s true organizational impact is clearer, investments in the right data readiness will be made.
