AI, LLMs, and finance: what UH Hilo's Wei Wei is building-and why it matters
Wei Wei, Assistant Professor of Finance at the University of Hawaiʻi at Hilo, is centering her research on how artificial intelligence, large language models (LLMs), and machine learning can improve financial analysis. Her focus is practical: use AI to detect risk, interpret complex disclosures, and anticipate market reactions-then apply those same tools to test what actually works.
Wei holds a master of business in finance from the University of Minnesota's Carlson School of Management and a PhD in finance from the University of North Carolina at Charlotte. She is a certified Financial Risk Manager (FRM) and has pursued the Chartered Financial Analyst (CFA) program, bringing a strong risk and investments lens to her work. Before UH Hilo (joining in 2025), she lectured at UNC Charlotte's Belk College of Business.
Why finance teams should care
Markets are text- and audio-driven. Earnings releases, guidance updates, 10-Ks, and earnings calls move prices because they change expectations. Wei's projects ask a blunt question: can AI improve how we write, read, and act on those signals?
Working papers with direct implications
How to write returns: a counterfactual editor for earnings guidance and press releases
Investors increasingly analyze disclosures with LLMs. This study goes beyond prediction to prescription: "How should managers write earnings disclosures to improve investor understanding?" Wei develops a counterfactual CAR editor that proposes fact-preserving edits to press releases and guidance, then estimates the change in abnormal returns using LLMs (ChatGPT, Claude, Gemini) with uniform prompts. Her takeaway: precise edits that keep facts intact can surface the underlying information more clearly and shift how the market reads the text.
Finance takeaway: Treat disclosure language as a testable variable. Build an internal, compliance-reviewed process to A/B test fact-preserving edits before release, keep an audit trail, and align with Reg FD.
Audio context: the causal effect of CEO communication
With co-author Patrick S. Smith, Wei examines how a CEO's ability to communicate-separate from their management style-affects analyst forecasts and firm performance. Using NLP to abstract "arousal" levels from earnings call Q&A, the study finds that communication quality is a causal channel impacting outcomes.
Finance takeaway: Track CEO voice metrics (arousal, clarity, pacing) in addition to sentiment. Coach for Q&A performance, not just scripted remarks.
Racial diversity and firm commitments
Co-authored with Al (Aloke) Ghosh and under review at Accounting and Business Research, this work analyzes race-related discussions in earnings calls and 10-K filings following the death of George Floyd. Drivers of increased attention: corporate integrity, strong ESG profiles, and pre-existing diversity policies. A difference-in-differences design shows firms that discuss race-related matters deliver measurable improvements in DEI performance, with favorable reactions from analysts, rating agencies, and-at times-investors.
Finance takeaway: Markets reward credible progress. Pair DEI communications with clear initiatives, metrics, and follow-through to avoid "diversity washing."
Short interest and financial reporting misstatements
Contrary to prior claims of an "inverted U" in short interest around misstatement announcements, Wei's broader sample finds little evidence of a pronounced curve. What does show up: elevated short-selling before and after announcements, largely tracking public negative signals like weak performance, legal issues, downgrades, potential delistings, and leadership departures.
Finance takeaway: Treat short-interest spikes as prompts to scan public catalysts and scenario-test liquidity and covenant headroom. The signal often reflects public data interpreted fast-not secret information.
Practical steps for CFOs, IR, and risk teams
- Set up an LLM-powered disclosure review. Test fact-preserving edits for clarity and potential market impact. Involve legal, keep version control, and align with your disclosure committee.
- Instrument your earnings calls. Measure arousal and clarity in Q&A, not just sentiment in prepared remarks. Coach executives with data.
- Tie DEI communications to actions. Define metrics, publish progress, and reconcile with ESG scores and internal policies.
- Monitor short interest alongside public catalysts. Build watchlists for legal items, downgrades, turnover, and guidance shifts; rehearse response plans.
- Invest in AI/NLP literacy for finance. Focus on data quality, prompt discipline, model risk management, and ethical use. For curated resources, see AI tools for finance.
Teaching the next wave of finance talent
Wei brings live research into the classroom, giving students hands-on exposure to data, analytical tools, and collaborative problem-solving. "This type of work helps automate routine analysis while enhancing accuracy and transparency," she says. She emphasizes ethics and social justice so graduates can use financial skills responsibly in their communities-and at your firm.
Background snapshot
- Assistant Professor of Finance, University of Hawaiʻi at Hilo
- Master of business in finance, University of Minnesota (Carlson)
- PhD in Finance, UNC Charlotte
- FRM certified; pursued the CFA program
- Former lecturer, UNC Charlotte Belk College of Business
Wei's throughline is simple: use AI to make disclosures clearer, calls more informative, and risk signals cleaner-then measure the effect. That is actionable research finance teams can put to work now.
Your membership also unlocks:
