Tag
This paper proposes a two-stage adapter that embeds foundation model predictions into a multinomial logit model, preserving economic properties like cost monotonicity and interpretable willingness-to-pay while improving accuracy by up to 12.8 percentage points.
This paper develops a formal theory of cognitive debt, where using AI as a substitute for first-principles reasoning builds up unverified obligations that lead to systemic fragility and a cognitive Minsky moment, showing that decentralized equilibrium over-adopts substitutive AI without accounting for externalities.
This paper formulates LLM inference budget allocation as a constrained optimization problem, proposing CLEAR to reallocate resources from low-utility queries to those near emergence thresholds, achieving up to 3× accuracy improvement under tight budgets.
This paper develops an economic model combining scaling laws with microeconomic theory to analyze profit-optimal training of large language models, considering trade-offs between model quality, training costs, and hardware efficiency.