ChatGPT Outpredicts Wall Street

Hi, I'm Matt. Welcome to AI Street, where AI meets Wall Street. Every Thursday, I share news, analysis, and expert interviews.

The Rundown

• Big Picture: Macro AI

• ChatGPT Bests Market: CapEx Study

• Interview: Bain’s Richard Lichtenstein on AI in PE

• Early Days: The 1996 Era of AI

• Hedge Fund Taps New Type of Data Whiz

• AI Safety Bill Shot Down

• 🇺🇲 Matt in Milan 🇮🇹 — Fintech Summit Edition

BIG PICTURE

• California Governor Gavin Newsom vetoed an AI safety bill. (See more below)

• OpenAI raised $6.6 billion in venture capital — in one of the largest funding rounds ever — giving it a market value of about $157 billion. (Associated Press)

• Microsoft will invest €4.3 billion ($4.75 billion) in Italy over two years to expand its cloud and AI infrastructure, making its Northern Italy cloud region one of Europe's largest. (WSJ)

• Accenture and NVIDIA have formed a new business group to accelerate enterprise AI adoption by helping companies implement generative and agentic AI solutions across various industries. (Accenture)

RESEARCH
ChatGPT Bests Wall Street in Predicting Corporate Spending: Study

ChatGPT outperformed traditional methods in predicting future company expenditures after analyzing over 74,000 earnings call transcripts, according to a study from researchers at Georgia State University and the University of Chicago.

“It was a bit of a surprise at how much of a leap ChatGPT was from earlier models,” said Manish Jha, a professor at Georgia State and one of the study's authors.

ChatGPT effectively assessed a firm's investment expectations, outperforming market predictions by forecasting both short-term and long-term declines associated with increased investments. The market typically fails to fully account for the long-term challenges and risks associated with such large capital expenditures.

Jha is expanding his research. He’s using ChatGPT to analyze over 120,000 conference call transcripts to extract managerial expectations about economic factors. The resulting "AI Economy Score" predicts future economic indicators such as GDP growth, production, and employment for up to 4 to 10 quarters, outperforming existing measures including survey forecasts.

The AI Economy Score paper is expected to be published next month.

INTERVIEW
AI Adoption in Private Equity

Richard Lichtenstein, Expert Partner and Chief Private Equity Data Officer at Bain & Company, shares his insights on AI adoption in private equity and financial services. In this interview, he discusses the current state of AI implementation, challenges, and future prospects in the industry.

This interview has been edited for clarity and length.

What are the main challenges in AI adoption for PE firms?

One of the biggest challenges is the limited in-house expertise. Most investors do not have armies of engineers working at the fund level. Even PE funds that we would consider to be large might only have one to three data scientists working at the whole fund. They're just not in a position to start engineering complicated solutions to problems because they don't want to maintain them. That's a blocker.

Another challenge is the lack of trustworthy vendor solutions. There are a lot of use cases that I can talk about, but there really aren't many companies who've actually built plug-and-play solutions that deliver those use cases in a consistent way yet.

How does AI adoption in PE compare to other areas of financial services?

If you think about other types of investors besides PE, groups like sovereign wealth funds are all over this. They're even more enthusiastic than many PE investors because they have investments across a wide variety of companies and asset classes. They see a lot more data than many PE investors do because they just have more investments. They're sitting on a lot of data, and they believe that if they can act on that data, it'll be exciting. Also, they have fairly deep pockets, so they can invest money in things that will make them better.

Hedge funds are a bit different. There's probably appetite for hedge funds to use this to detect market signals. For hedge funds trying to find things quickly, using AI to detect market signals is sort of a no-brainer, and the most sophisticated ones are probably all doing that. If you're a value investor, your behavior is more similar to a PE fund, so your adoption is probably similar as well. You might be using some productivity stuff, but you're also starting to invest with a thesis of which companies could be interesting for AI or which companies could be the winning picks and shovels for AI.

Read the full interview here

INTERVIEWS
ICYMI: From our “5 Minutes with” series:

• Snowflake’s Jonathan Regenstein on AI building novel datasets

• Skadden’s Dan Michael on the SEC’s AI stance

• Stardog’s Matt Lucas on Hallucination-Free AI

• Celent’s Monica Summerville on AI Adoption in capital markets

• Aveni's Joseph Twigg on building a Finance LLM

• Persado’s Assaf Baciu on tailored AI marketing at banks

• Professor Alejandro Lopez- Lira on AI-driven stock predictions

ADOPTION
We're Only in 1996 of AI Era: BofA

• AI could boost S&P operating margins by 200 basis points over 5 years, saving ~$55 billion annually

• 2024 expected for enterprise AI ROI determination; 2025 for widespread adoption

• Current AI state compared to internet in 1996; long-term impacts likely underestimated

• Enterprise AI applications are likely to use small-language models and retrieval-augmented generation for specific use cases, rather than broad models like ChatGPT.

• From AI: From evolution to revolution?

UK Financial Firms Double AI Investment: Lloyds

UK financial institutions have dramatically increased their AI investment in the past year, a Lloyds Bank survey reveals.

• 63% of firms now invest in AI, up from 32% last year

• 81% view AI as a business opportunity

• Benefits already seen: improved productivity, competitive advantage, better customer insights

• Lloyds surveyed 100 senior decision-makers at banks, wealth and asset managers, insurers, and financial sponsors.

HEDGE FUNDS
AI Extracting New Insights for Hedge Funds

Systematica Investments, one of the largest female-led hedge funds, is using AI to create novel datasets, including sentiment analysis of executive and policymaker speeches. The team recently considered investing in a program that evaluates tone of voice, per Bloomberg.

“A data set 10 years ago was a time series,” said Leda Braga, founder and CEO of Systematica. “No longer. A data set is anything now, a collection of images, databases of things. What’s the X-axis of that? I don’t know.”

AI has led to an explosion of new data that requires even more specialized staff, including data scientists and AI experts. LLMs like ChatGPT can create novel datasets from unstructured data like emails and PDFs.

As Jonathan Regenstein told me last week: "There's a growing recognition that there are incredible insights sitting in the world of unstructured data." 

REGULATION
California Governor Vetoes Controversial AI Safety Bill

California Gov. Gavin Newsom vetoed an AI safety bill that would have held developers accountable for “severe harm.”

Companies developing powerful AI models would have been required to take reasonable care to ensure that their technologies wouldn’t cause mass casualties or property damage above $500 million.

The legislation was set to have major impact on the AI industry, given California's central role in the burgeoning tech industry and the absence of federal AI regulations.

The bill had sparked debate among tech leaders, politicians, and AI companies. Critics, including major tech firms and some Democrats like Nancy Pelosi, argued that the bill could stifle innovation and drive AI companies from the state. (Associated Press)

FUNDRAISING
Rogo raises $18 Million for AI Analyst

Rogo raised $18.5 million in Series A funding led by Khosla Ventures. The round brings Rogo's total funding to $26 million and includes participation from Mantis VC, Jack Altman (Sam’s brother), and Eric Schmidt.

Keith Rabois of Khosla Ventures joins Rogo's Board of Directors. The funding will support team growth, platform development, and customer partnerships.

Founded in 2021, Rogo fine-tunes large language models for financial applications. The platform has been adopted by about 25 financial firms and achieved multi-million-dollar annual recurring revenue, per a Oct. 1 press release.

CEO Gabriel Stengel, a former Lazard investment banker, said "We believe Generative AI is set to revolutionize knowledge work, especially in financial services."

There are lots of AI companies entering the AI-as-financial-analyst space. I put together a list of more than 40+ firms. Contact me if you’d like a copy: [email protected].

🇺🇸 Matt in Milan 🇮🇹 
Milan FinTech Summit

This week I have a “real” Matt in Milan post — as opposed to sharing a fairly obvious cultural observation about a New Yorker living in Milan — as I’ve been doing 🙂 

I’ll be moderating an AI panel at the Milan Fintech Summit next week.

Fortunately, it’s all in English as my Italian skills have already been surpassed by my two year old.

If you’re attending, come say hi. My session is on Oct. 9 at 11:40 a.m. at the Arena.

See you next week!