CONCEPT
Learning Debt
The analog of technical debt for the AI-assisted startup — the accumulated liability of <em>experiments conducted but not analyzed</em>, whose interest is the compounding cost of decisions made without information the unanalyzed experiments would have provided.
Learning debt is the accumulated cost of experiments whose results have been collected but not processed into understanding. In the pre-AI regime, learning debt accumulated slowly because production pace allowed analysis to keep up. In the AI-assisted regime, it can accumulate rapidly: a startup shipping ten features per month but analyzing only three has accumulated debt on seven. The features are in production, generating data. But the data is not being processed. The startup knows what was built; it does not know whether what was built is working. Learning debt should be tracked as a liability on the innovation accounting balance sheet, with growth rate serving as a warning signal that production is outpacing learning.
In The You On AI Field Guide
The parallel to technical debt is deliberate and precise. Technical debt is the accumulated cost of code that works now but will cost more to maintain later; learning debt is the accumulated cost of data that exists now but