The Literacy Gap
Summary: The gap between how accessible AI tools have become and how little the people using them, including technology teams, actually understand about what they’re doing. Fills a gap flagged earlier in duties of ai interview questions as not yet having its own page.
Sources: sources/The Skills of AI versus The Duties of AI.md (a work in progress, per Paul, 2026-07-08); Templates/Conversation Structure Template.md (Segment 2 background notes)
Last updated: 2026-07-09
The core observation
Paul’s own framing: “while the skills have become simple and accessible for mass use, literacy remains extremely low.” (source: The Skills of AI versus The Duties of AI.md) The barrier to touching an AI tool has fallen; the barrier to understanding what it’s actually doing, and when to distrust it, has not moved.
The Amazon River analogy
Paul’s own vivid version of the idea: “it’s like we went up the Amazon River and found people still hunting with bow and arrow and started handing out machine guns. We told them that failure to adopt the machine gun meant extinction, but we did not teach about guns or how to operate the machine gun. And life did not improve, but bullets are flying everywhere. A dark vision but not inaccurate.” (source: The Skills of AI versus The Duties of AI.md)
He also names the incentive behind the mismatch directly: “once the huge investments in foundation models were made, the builders were desperate to create demand. Marketing fear of a future without deeply integrated dependence on these models became the standard message.” (source: The Skills of AI versus The Duties of AI.md) The urgency being sold and the literacy needed to use the tool safely are not the same thing, and only one of them gets marketed.
”The humans are hallucinating more than the AI”
Paul’s own compression of the same dynamic (source: conversation with Paul, 2026-07-09): low literacy and a genuine weakness in explaining how AI actually works, combined with a strong mass-market message that integrating AI into daily life is nearly mandatory for future success, produces the same fear-mongering already named above as the “standard message.” That fear-mongering is what drives the swing between over-enthusiasm and paralysis described below, people acting on belief rather than understanding, which is its own kind of hallucination, arguably a more consequential one than anything a model produces. Goes hand in hand with the Amazon River / machine gun analogy above.
Not just a business-side problem
The background notes for a planned interview on this exact topic (source: Templates/Conversation Structure Template.md) make a point Paul found genuinely surprising: he expected his business partners to need grounding in how AI works, but his technology partners needed just as much. Leadership teams, in this framing, oscillate between paralyzing anxiety and irrational euphoria, neither a useful posture, and the foundational gap underneath both isn’t AI-specific: most executive teams don’t have a clear picture of how data becomes insight becomes a decision, and AI layered on top compounds a gap that was already there.
The specific version: analytics literacy and IT overconfidence
Paul narrowed the focus directly (source: conversation with Paul, 2026-07-08): the gap he means here is analytics literacy specifically, not general AI awareness. Its sharpest version shows up in traditional IT leaders who deploy AI confidently without understanding the duties that come with it, in particular regulatory requirements. This gives a concrete failure mode to the duties of ai’s Regulatory Compliance duty: a leader can be fluent enough in the tooling to ship a model and still have no grounding in what that model now obligates the organization to prove or maintain once it’s live. The overconfidence itself is the dangerous part, it isn’t hesitation or a skills gap, it’s the mismatch between how easy the tool was to deploy and how little the deployer understood about what deploying it actually commits the organization to.