Pharma AI glossary
Jagged frontier
The defining metaphor for AI capability, from the 758-consultant Harvard/BCG field experiment: the boundary of what AI does well is uneven and invisible, +12% task completion, +25% speed, +40% quality inside it; −19 percentage points accuracy just outside it.
Definition
The defining metaphor for AI capability, from the 758-consultant Harvard/BCG field experiment: the boundary of what AI does well is uneven and invisible, +12% task completion, +25% speed, +40% quality inside it; −19 percentage points accuracy just outside it. Corollaries: capability must be tested empirically per task (not inferred from adjacent tasks), and user intuition about where the edge sits is unreliable. Arguably the strongest single argument for why AI literacy programs exist.
Not a smooth capability curve
The jagged frontier is the finding that AI is strong on some tasks and weaker than humans on adjacent tasks that look identical, so users cannot feel the edge. BCG’s 758-consultant experiment showed large gains inside the frontier and a 19-point increase in error rates outside it when people relied on AI without independent checks.
Pharma translation
An agent may accelerate investigation or batch-record review inside a bounded workflow yet still produce unsupported citations if retrieval or review is weak. The same applies to CAPA summaries and change-control packages. Literacy is mapping task-level limits, enforcing harness stop points before QMS write, and routing edge cases to humans.
Frequently asked questions
What is falling asleep at the wheel?
The BCG-named pattern where better AI increases deference, including where the model is wrong. Training must reward verification, not speed alone.
How do teams map their own frontier?
Pilot with scored tasks across the workflows you intend to automate (investigations, batch review, inspection assist, document search), independent review, and explicit outside-frontier examples. Module 1 of the Literacy Center uses frontier-sort exercises for this.
Related terms
Benchmark your AI literacy
Free 20-question assessment, eight modules, and instant scoring, built for pharma quality and manufacturing teams.
Open the Literacy CenterBrowse all terms
See the full 71-term glossary index or search interactively in the Literacy Center.
Back to glossary hub →Educational content only, not legal or regulatory advice. Regulatory guidance cited here includes drafts (FDA AI credibility guidance; EU GMP Annex 22) as of August 2026; verify against final texts before relying on them in submissions. Company-reported figures (Merck CSR timings, Sanofi results) are labeled where used. MIT's ~95% pilot figure carries its own caveat (~150 interviews, contested definitions, not peer-reviewed). Re-check sources on module finalization.