Why does Bank of America start with the client need instead of the model?

Hari Gopalkrishnan said BofA asks whether AI actually answers the problem, because rushing to AI where deterministic models suffice is one of the biggest mistakes.

What results did S&P Global's work with a Tier 1 bank produce?

S&P Global helped the bank reduce time to market by roughly sixfold and improve accuracy in work drawing on multiple data sets from about 60% to 98%.

What is the risk of a chatbot that doesn't serve certain accents?

Gopalkrishnan said such gaps create customer, compliance and reputational risks, not merely a technology problem.

AI Safety & AlignmentAI Business & IndustryFortune AI4m agoOct 2, 2026, 22:00 JST

BofA, S&P Global: AI wins need governance, data

Send on LINE

3 Key Points

What happened
At Fortune's AIQ Summit, Bank of America's Hari Gopalkrishnan said the bank evaluates AI across 16 risk dimensions, and S&P Global's Sally Moore said traceability to source data is essential.
Why it matters
Gopalkrishnan warned the costliest error may be funding AI where simpler models suffice, so governance and provenance, not model choice, appear to drive trustworthy returns.
What to watch
S&P Global's cited gain came from one Tier 1 bank engagement, so the test is whether the roughly sixfold speedup and 98% accuracy hold across other clients.

WHO IT HITSChief data and compliance officers at banks and other regulated firms face pressure to show AI decisions can be traced to source data, while technology leaders must justify AI spending against simpler alternatives.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

The panel remarks arrive alongside Fortune and ServiceNow's release this week of the Fortune AIQ 75, an annual ranking of Fortune 500 companies generating measurable impact from AI. That timing frames the discussion less as a product race and more as a discipline question: how to prove an AI decision can be trusted before it is deployed.

Gopalkrishnan's list of 16 risk dimensions, spanning privacy, bias, workforce implications and intellectual property, sits at one end of that discipline. Moore's emphasis on sourcing the original IP behind AI outputs sits at the other. Together they describe a workflow where guardrails and provenance are continuous, not one-time checks.

A separate Bain & Company report in the same edition estimates AI could shift $4.7 trillion in global corporate profits from 2025 through 2035, with about 75% coming from innovation and competitive redistribution rather than productivity gains alone. The stakes for regulated firms thus hinge on whether governance and traceability can be operationalized at scale—an open question given the single Tier 1 bank example cited.

FAQ

Why does Bank of America start with the client need instead of the model?: Hari Gopalkrishnan said BofA asks whether AI actually answers the problem, because rushing to AI where deterministic models suffice is one of the biggest mistakes.
What results did S&P Global's work with a Tier 1 bank produce?: S&P Global helped the bank reduce time to market by roughly sixfold and improve accuracy in work drawing on multiple data sets from about 60% to 98%.
What is the risk of a chatbot that doesn't serve certain accents?: Gopalkrishnan said such gaps create customer, compliance and reputational risks, not merely a technology problem.

Fortune AIRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.