Scott Shimp, VP of Solutions at Data Society, says some organizations avoid the word “literacy” in AI training because it sounds like a grade: literate or not, fluent or not. His fix is a different frame, not a new word. Data and AI are a language, and the goal is basic shared understanding, not a pass/fail test. Industry data on how AI training actually lands backs up why that distinction matters.
The scenario Scott Shimp describes
Shimp names the discomfort directly, then reframes it:
“Some organizations might be hesitant around the term literacy, because people are sensitive around that term. But fundamentally, if you think about it as a language, that’s what we’re talking about here. So it’s not about is someone literate or not literate, fluent or not fluent. We’re trying to establish that basic understanding of a language around data and AI.”
Scott Shimp, Vice President, Solutions, Data Society
Why the word itself trips people up
“Literacy” carries baggage “understanding” doesn’t. It implies a threshold: you either have it or you don’t, and not having it reads as a deficiency rather than a stage. That’s part of why a live debate has grown up around whether organizations should even use “AI literacy” at all, with some L&D leaders pushing “AI fluency” instead on the theory that fluency sounds like a spectrum you move along rather than a line you’re on the wrong side of. Shimp’s framing sidesteps the word fight entirely: whatever you call it, the actual goal is a shared vocabulary, not a certification.
What the training data shows happening without that shared vocabulary
A 2026 industry survey on data and AI literacy found 88% of enterprise leaders call basic data literacy essential to daily work, and 72% say the same about AI literacy. Despite that, 60% report an active data skills gap and 59% report an AI skills gap in their own organization. Seventy-seven percent offer some form of AI training, but only 35% have a mature, organization-wide AI literacy program, and only 42% provide foundational data literacy training at scale. Training already exists. The gap is that most of it doesn’t build the basic, shared understanding Shimp is describing. The report points to passive, video-based courses with no hands-on practice and no role-specific relevance as the recurring complaint.
Why “a language” is the more useful frame
Nobody expects fluency the first week of learning a language, and nobody treats a beginner speaker as a failure for not being fluent yet. Everyone’s expected to be somewhere on a continuum, and the goal is communication, not a diploma. Applied to AI, that reframing changes what a training program is actually for: not sorting people into literate and illiterate, but giving everyone in an organization enough shared vocabulary that they can talk about data and AI, evaluate a claim about either, and know when to ask for help.
Frequently asked questions
Why are some organizations hesitant to use the term “AI literacy”?
Because it implies a pass/fail threshold: literate or not, according to Scott Shimp. That framing can feel more like a judgment than a skill people are building.
What does Scott Shimp mean by treating AI literacy as a language?
That the goal is basic, shared understanding and vocabulary around data and AI, not certifying who is or isn’t “literate” or “fluent.”
How big is the AI skills gap in most organizations?
Substantial. A 2026 industry survey found 59% of organizations report an AI skills gap despite 72% of leaders calling AI literacy important, and only 35% have a mature, organization-wide AI literacy program.
Is “AI fluency” a better term than “AI literacy”?
Opinions differ. Some L&D leaders prefer “fluency” because it reads as a spectrum rather than a threshold, which is close to the same problem Shimp is naming with the word “literacy” itself.
Why does most AI training fail to close the skills gap?
Industry data points to passive, video-based training without hands-on practice or role-specific relevance, not a lack of investment. Seventy-seven percent of organizations offer some AI training already.
Next steps
Whatever your organization calls it, literacy or fluency, the test is whether people across teams can actually talk to each other about data and AI. Data Society’s AI upskilling programs are built around hands-on, role-specific practice instead of passive video courses, and the training catalog has the current course lineup.
