AI & Technology
More AI Drafts, Fewer Real Decisions
Why does a senior analyst at a major consulting firm spend more time polishing AI‑generated slides than choosing the next client strategy?
2026-09-221 min read
The paradox of “more AI drafts, fewer real decisions” emerges when the speed of generation masks the true bottleneck: judgment. An AI can spin a polished deck in minutes, but each draft still requires a human to decide which insight matters, which hypothesis to test, and which recommendation to champion.
The shortcut works because the model supplies a veneer of completeness, letting professionals skip the uncomfortable step of framing the problem themselves. In a recent internal meeting, a lead consultant opened a freshly generated market‑size slide, then spent the entire session debating the color palette and font size while the strategic question of market entry remained untouched.
The AI’s convenience created a silent feedback loop: the more polished the output, the less the team felt compelled to interrogate the underlying assumptions, and the more the decision‑making process stalled behind a wall of aesthetics. The hidden cost is not lost productivity but eroded expertise, as the habit of “polish first, think later” rewires the brain to prioritize surface over substance.
Key insights
AI drafts often satisfy the “look good” metric while sidestepping the “what now?” metric.
The habit of polishing before questioning rewires teams to equate output speed with strategic clarity.
Why it matters
Ignoring this drift leaves critical strategic choices to be made by consensus on a document that never asked the right questions.
Over time, the team’s collective ability to generate original hypotheses fades, making the organization vulnerable to competitors who still value deep, unautomated thinking.
Use this tomorrow
1Open the most recent AI‑generated proposal you received and count how many bullet points address a concrete business decision versus how many merely describe data visualizations.
2In the same document, highlight any paragraph that begins with “We recommend” and ask yourself whether the recommendation stems from a data‑driven insight or from the AI’s phrasing; note the result.
Go deeper
The phenomenon traces back to cognitive‑load theory: when mental effort is offloaded to a tool, the brain reallocates resources to lower‑level tasks like formatting. Researchers on human‑AI interaction have observed that this shift reduces the time spent on higher‑order reasoning, even when the tool’s output is flawless.
However, the effect is not uniform; teams that embed a “question‑first” checkpoint before AI generation preserve their analytical muscle. In fields where regulatory scrutiny demands documented rationale, skipping that step can trigger compliance failures far more costly than any time saved.
