Mechanism Design for Large Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Duetting, Paul, Mirrokni, Vahab, Leme, Renato Paes, Xu, Haifeng, Zuo, Song
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929405198073856
author Duetting, Paul
Mirrokni, Vahab
Leme, Renato Paes
Xu, Haifeng
Zuo, Song
author_facet Duetting, Paul
Mirrokni, Vahab
Leme, Renato Paes
Xu, Haifeng
Zuo, Song
contents We investigate auction mechanisms for AI-generated content, focusing on applications like ad creative generation. In our model, agents' preferences over stochastically generated content are encoded as large language models (LLMs). We propose an auction format that operates on a token-by-token basis, and allows LLM agents to influence content creation through single dimensional bids. We formulate two desirable incentive properties and prove their equivalence to a monotonicity condition on output aggregation. This equivalence enables a second-price rule design, even absent explicit agent valuation functions. Our design is supported by demonstrations on a publicly available LLM.
format Preprint
id arxiv_https___arxiv_org_abs_2310_10826
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Mechanism Design for Large Language Models
Duetting, Paul
Mirrokni, Vahab
Leme, Renato Paes
Xu, Haifeng
Zuo, Song
Computer Science and Game Theory
Theoretical Economics
We investigate auction mechanisms for AI-generated content, focusing on applications like ad creative generation. In our model, agents' preferences over stochastically generated content are encoded as large language models (LLMs). We propose an auction format that operates on a token-by-token basis, and allows LLM agents to influence content creation through single dimensional bids. We formulate two desirable incentive properties and prove their equivalence to a monotonicity condition on output aggregation. This equivalence enables a second-price rule design, even absent explicit agent valuation functions. Our design is supported by demonstrations on a publicly available LLM.
title Mechanism Design for Large Language Models
topic Computer Science and Game Theory
Theoretical Economics
url https://arxiv.org/abs/2310.10826