Practical Principles for AI Cost and Compute Accounting
Policymakers are increasingly using development cost and compute as proxies for AI model capabilities and risks. Recent laws have introduced regulatory requirements that are contingent on specific thresholds. However, technical ambiguities in how to perform this accounting could create loopholes that undermine regulatory effectiveness. This paper proposes seven principles for designing practical AI cost and compute accounting standards that (1) reduce opportunities for strategic gaming, (2) avoid disincentivizing responsible risk mitigation, and (3) enable consistent implementation across companies and jurisdictions.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Goal-based portfolio selection with mental accounting
We present a continuous-time portfolio selection framework that reflects goal-based investment principles and mental accounting behavior. In this framework, an investor with multiple investment goals constructs separate …
Towards Resource-Efficient LLMs: End-to-End Energy Accounting of Distillation Pipelines
The rise in deployment of large language models has driven a surge in GPU demand and datacenter scaling, raising concerns about electricity use, grid stress, and the impacts of modern AI workloads. Distillation is often …
Knowledge DistillationTaxation of a GMWB Variable Annuity in a Stochastic Interest Rate Model
Modeling taxation of Variable Annuities has been frequently neglected but accounting for it can significantly improve the explanation of the withdrawal dynamics and lead to a better modeling of the financial cost of thes…
Efficient privacy loss accounting for subsampling and random allocation
We consider the privacy amplification properties of a sampling scheme in which a user's data isused in $k$ steps chosen randomly and uniformly from a sequence (or set) of $t$ steps. This sampling scheme has been recently…
Anchorless Diversification for Parallel LLM Ideation
LLMs are increasingly used to generate candidate-idea pools for creative tasks where broad exploration is valuable. Parallel inference can be attractive in this setting when it broadens the pool while retaining quality a…