ChatGPT and Claude struggle to structure municipal bond portfolio

5 articles · Updated · Barron's · May 8

In a test to build a diversified $1 million muni portfolio, both chatbots offered broad laddering advice, while ChatGPT also produced incorrect CUSIP numbers.
Claude said it could suggest issuers, maturities and credit criteria but not specific bonds, directing investors to EMMA or brokers such as Fidelity and Charles Schwab.
The exercise underscored the municipal market's issuer-by-issuer complexity, suggesting low-cost muni ETFs may suit many individuals, while human advisers can better tailor portfolios and assess call features and local risks.

Can AI ever master the bond market, or will human expertise always be essential for complex investments?

If an AI financial advisor loses your money, who is to blame?