William Wyatt
> PhD Economics > MS Mathematics > BS Physics

William Wyatt

Studying how LLMs reason, decide, and take risks

I run behavioral experiments on language models using game theory and causal inference to figure out how prompts shape the way AI makes decisions. My work connects economics, alignment, and empirical AI research.

10 Publications
3 Degrees
14 Years Coding
6 Languages

Research

LLM behavioral alignment through the lens of economics

Job Market Paper

Participation or Observation: How Prompts Control LLM Reasoning

Empirical study across 283 prompt permutations identifying which phrases cause LLMs to embody scenarios versus observe them, and how this shifts strategic behavior.

Under Review

The Role of Responsibility in AI's Strategic Risk-Taking

15,000+ observations showing LLMs are 73.2% more likely than humans to take risks when decisions affect others.

Working Paper

Behavioral Effects of Personas and Demographics

When LLMs play behavioral games, their output changes based on the persona they take on. Mapping which demographic dimensions shift AI strategic choices.

2026
Working Paper

Measurable Effects from Emergency Zero Bail in California

Quasi-experimental analysis of zero bail policy reform using RDD and DiD to measure real-world outcomes of criminal justice interventions.

2026
wil@arch:~/projects$ ls -la

Projects

Tools and systems I've built

TrueNorth

Accepted, 2025

Conversational agentic RAG using PERMA+4 framework to optimize long-term STEM engagement. LLM orchestration + Discord.

PythonLLMRAG
View →

ApexLabel

Annotation Tool

Annotation software for YOLO training using SAM. Bootstraps object detection with CLIP transformer embeddings.

PythonYOLOSAM
View →

Market Simulation

Economics

Agent-based simulation of free market dynamics. Models competitive equilibrium, price discovery, and emergent behavior.

PythonEcon
View →

Geocoder

Computational Justice Lab

150M police stops at 280 req/s. Python + Rust. Saved $200k/yr in API costs for criminal justice research.

PythonRust

MTurk Server

Research Infra

Web interface for managing behavioral experiments. Annotation, survey, and data collection pipeline for MTurk and students.

PythonFlask
View →

Fireside

LLM Game

Conversational game built on LLMs. Interactive dialogue exploring narrative generation and player-AI interaction.

PythonLLM
View →
Older builds

IOTA Workshop

Autonomous equipment rental on distributed ledgers. Raspberry Pi + IOTA. Won IOT2Tangle Hackathon.

View →

Bot Orchestration

Multi-agent bot swarm with C&C server. Recorded human mouse movements for realism. Docker isolation, custom scripting DSL.

Webcrawling CTF

Multiplayer educational capture-the-flag teaching web crawling. Matter.js WebGL, Flask backend.

View →
CH1 60Hz

Character Sheet

William Wyatt at UCSC

Research Interests

Skills

🐍PythonLv.14
14yr / Master
C++Lv.14
14yr / Expert
🦀RustLv.4
4yr / Proficient
🗄SQLLv.8
8yr / Expert
📊STATALv.5
5yr / Proficient
📈RLv.3
3yr / Journeyman
🔥PyTorchLv.4
4yr / Expert
🐧LinuxLv.15
15yr / Master
🧠LLMsLv.4
4yr / Master
🔬Causal Inf.Lv.6
6yr / Master
CATALOG
NO. 10
WYATT, W. — COLLECTED WORKS
2019–2026

Publications

Economics & AI

The Role of Responsibility in AI's Strategic Risk-TakingUnder Review, 2026
Participation or Observation: How Prompts Control LLM ReasoningJob Market Paper, 2026
TrueNorth: Conversational Agentic RAG to Optimize STEM EngagementAccepted, 2025
Private Security Confounds Estimates of Public Police and CrimeSpringer, 2023

Physics (UCSC / SLAC)

Get in touch

William.Wyatt@cgu.edu