Surjit Chana says his first patent, two years after graduating in physics, was for an early AI system called an expert system.
“my first patent two years after graduating physics was an early version of AI called Expert Systems.”
Person
Surjit Chana is a technology executive who appeared on the American Banker Podcast to discuss what a moral architecture for AI in banking would look like. He covered bias in lending, fairer credit decisions, jobs, and the energy cost of AI.
Chana says he ran IBM's roughly $1 billion supercomputing business and its roughly $5 billion mid-market business. He told the podcast that Zest AI's clients saw about a 25% rise in approval rates while holding risk constant. He also says he is finishing a white paper on the framework, which he plans to open source.
1 episodes · 23 quoted claims
Surjit Chana says his first patent, two years after graduating in physics, was for an early AI system called an expert system.
“my first patent two years after graduating physics was an early version of AI called Expert Systems.”
Surjit Chana says he ran IBM's roughly $1 billion supercomputing business and its roughly $5 billion mid-market business.
“I had the opportunity to run IBM's supercomputing business, which is about $1 billion, IBM's mid-market business, about $5 billion”
Surjit Chana recalls that IBM released an expert-system product in 1992.
“IBM, I think, released a product in 1992, if I recall correctly.”
Surjit Chana joined the Harvard Advanced Leadership Initiative Fellowship, a program that helps senior leaders apply their skills to societal issues.
“I came across the Harvard Advanced Leadership Initiative Fellowship Program that helps senior leaders to do exactly that”
Surjit Chana cites a 2025 Government Accountability Office report as confirming that AI-driven bias in lending is real.
“The 2025 Government Accountability Office report confirmed this is not theoretical.”
Surjit Chana says there are 1.3 billion unbanked adults worldwide and millions of credit-invisible Americans.
“There are 1.3 billion unbanked adults globally and millions of credit-invisible Americans.”
Chana cites a World Economic Forum projection that 92 million jobs will be displaced and 170 million created.
“92 million will be displaced and 170 million jobs will be created, are often cited reassuringly.”
Surjit Chana says data centers consume roughly 4.4% of all U.S. electricity.
“Data centers now consume roughly 4.4 % of all U.S. electricity, and the AI workloads are accelerating that figure.”
Zest AI is a technology company founded in Burbank, California, with a mission to broaden access to responsible lending.
“They're a technology company founded in Burbank, California, where their mission was to actually broaden access to responsible lending.”
Chana says Zest AI has worked with about 180 banks and credit unions.
“They've been working with institutions, actually about 180 banks and credit unions”
Chana says Zest AI's clients saw about a 25% rise in approval rates while holding risk constant.
“They saw an increase of about 25 % in approval rates while holding risk constant.”
Chana says Verity Credit Union ran AI underwriting with 300 variables rather than the traditional 15.
“When they ran the model with AI underwriting using 300 variables versus the traditional 15”
Chana says Verity Credit Union's AI model raised approvals by 177% for African-American applicants, 271% for people over 62 and 194% for women.
“African-American approvals increased 177%. Approvals for individuals over 62 increased 271%. Approvals for women increased 194%.”
Chana predicts that within three years a third of bank work could be fully automated by AI.
“I am thinking within three years that we're going to start to see those sort of numbers.”
Chana says training a single large language model can use enough electricity to power over 100 average homes for a year.
“trading a single large language model can consume enough electricity to power over 100 average homes for a year.”
Chana says he presented the framework to about 40 bank CEOs and 30 board members at the GABV annual meeting.
“I had the opportunity just a couple of weeks ago to present to about 40 of those bank CEOs and about 30 board members at the GABV annual meeting.”
Chana cites $70 million in combined penalties paid by Apple and Goldman Sachs in 2024 as an example of the cost of noncompliance.
“the $70 million in combined penalties Apple and Goldman Sachs paid in 2024 is a prominent example.”
Chana says Global Alliance for Banking on Values research found banks with strong ESG practices outperformed conventional peers.
“Research from the Global Alliance for Banking Values found that banks with strong ESG practices outperformed their conventional peers.”
Beneficial State Bank is a community bank on the U.S. West Coast with branches in California, Oregon and Washington.
“We have branches in California, Oregon, and Washington.”
Beneficial State Bank's board approved a comprehensive AI strategy covering 2026 to 2029 in January.
“In January, our board approved a comprehensive AI strategy for 2026 and 2029, designed to be exactly what I've been describing”
Beneficial State Bank says no AI capability is deployed without enterprise risk management validation of risk tiering, explainability, data usage and compliance alignment.
“No AI capability is deployed without ERM validation of risk tiering, explainability, data usage, and compliance alignment.”
Beneficial State Bank targets 5% to 10% productivity improvements through AI augmentation rather than workforce reduction.
“targets 5 % to 10 % productivity improvements through augmentation, not workforce reduction.”
Chana is finishing a white paper that goes two or three levels deeper into the moral architecture, which he plans to open source.
“I am wrapping up a thought leadership white paper which provides more depth to the architecture, two or three levels deeper”