The terms machine learning and artificial intelligence are often used interchangeably, but the former is actually an advanced subset of the latter. Just because something is artificially intelligent doesn’t necessarily mean it can learn.

Machine learning technology can adapt to different situations and learn as it goes — and the finance industry is taking advantage of these functions, implementing machine learning in all facets of finance.

Machine Learning in Finance

Machine learning is having a major impact in finance, from offering alternative credit reporting methods to speeding up underwriting. The finance industry is rapidly deploying machine learning to automate painstaking processes, open up better opportunities for loan seekers to get the loan they need and more.

Stephen Gossett contributed reporting to this story.


An overview of machine learning in fraud detection | Video: Siraj Raval

Machine Learning in Finance: The Push for Fairness

Today, machine learning impacts virtually every corner of finance. Hedge funds and investment firms use machine learning models, fed with vast amounts of traditional and alternative data, to help evaluate stocks and assets. Financial institutions and payments services alike use sophisticated machine learning to spot fraudulent activity. And financial experts have adopted AI to predict and manage risk.

But the adoption of machine learning remains notably fraught in one subcategory in particular: consumer finance. Many companies now use machine learning models when deciding whether to offer car loans, personal loans and mortgages. Optimists contend that more sophisticated models can lead to fairer loans than the notoriously problematic old-school FICO algorithm, but equity advocates counter that models trained on unfair patterns and biased data merely amplify existing loan discrimination. 

“We’re learning that these algorithms — not just in financial services, but across a bunch of domains — if left to their own devices, they do harmful things,” said Kareem Saleh, co-founder of FairPlay, a platform that measures algorithmic fairness in lending models.

“We’re learning that these algorithms — not just in financial services, but across a bunch of domains — if left to their own devices, they do harmful things.”

Several advancements could help mitigate machine learning lending bias, but one of the most important, according to Saleh, is increased self-auditing. Laws already prohibit discriminatory lending, but fluctuating levels of enforcement, plus the fact that regulators are still learning the intricacies of how AI and machine learning affect consumer underwriting, means accountability can be uneven. Also key, Saleh said, is taking a more holistic view of a borrower’s identity and information. For example, consistency of employment variables often discriminate against women who temporarily leave the workforce to raise or care for family, he noted.

At the same time, Saleh said that emerging fintech companies, if not always traditional lenders, are becoming more proactive about auditing models. Audits improve both company reputation and the bottom line — lenders ought to feel incentivized to accept borrowers who do repay with interest, regardless of what the FICO model warns. Plus the regulatory muscle might actually catch up in time: Washington, D.C.’s push for more transparency from companies that use AI and ML for high-impact decisions appears to be gathering momentum.

Upstart, an alt-data- and ML-based personal lender, is an interesting case study. The company agreed to share audit results with federal monitors in exchange for no regulatory action. The numbers were promising: 27 percent more approved applicants over a traditional model and 16 percent lower average annual percentage rates.

Still, such transparency remains rare, which means widespread, verified fairness remains a target, rather than a realization. 

“We are still in the first inning of this ballgame,” Saleh said.

That said, machine learning adoption in finance, whether consumer finance or investment funds, continues to expand on the ground. Here are 10 companies that help illustrate the landscape.


10 Companies Using Machine Learning in Finance

affirm machine learning applications finance
Image: Affirm


Location: San Francisco, California

What it does: Affirm is a payment service enabling consumers to finance items and pay for them over time. Consumers agree to the amount upfront so they know precisely how much they’re paying. Affirm is accepted at a large variety of retailers and businesses including Wayfair, Expedia, Peloton and Casper, making larger purchases more accessible and affordable.

Industry impact: Affirm can identify more credit-deserving consumers than traditional scoring systems through intelligent underwriting models, according to the company. These models use machine learning to more accurately assess ability to repay and fairly price risk at the point of sale. This helps reduce fraud rates and defaults while allowing users to have wider access to credit.


datavisor machine learning applications finance
Image: Datavisor


Location: Mountain View, California

What it does: Datavisor uses unsupervised machine learning to catalyze fraud detection. With unsupervised learning, no retraining is necessary for machines to detect new types of fraudulent activity. Datavisor’s technology combines graph analysis with clustering techniques to detect patterns in unlabeled data across billions of accounts.

Industry impact: Datavisor’s technology protects more than four billion accounts worldwide with its technology, according to the company’s site.

More on Fintech7 Examples of How IoT Is Improving Fintech


deserve machine learning applications finance
Image: Deserve


Location: Menlo Park, California

What it does: Deserve’s credit cards help young adults build their credit history. With a focus on students, the card also offers no annual fees and rewards users for making specific purchases. Deserve employs machine learning tools instead of traditional credit-worthiness sources to approve its cardholders.

Industry impact: In June, Deserve raised $50 million from investors, including Mastercard and Ally Ventures, to ramp up operations.


enova machine learning applications finance
Image: Enova


Location: Chicago, Illinois

What it does: Enova develops and provides a variety of financial products and services for businesses and individuals. The company’s brand, Enova Decisions, is used in multiple industries — including finance. The service helps companies gain more customers through machine learning models that provide personalized risk and credit analysis.

Industry impact: Enova Decisions has helped companies like online lender Headway Capital to automate decisions, assess default risk and provide customers with pre-qualified loans and pricing in real time.

RelatedRead More Fintech Stories


feedzai machine learning applications finance
Image: Feedzai


Location: San Mateo, California

What it does: Feedzai works with global finance companies, banks and retailers to provide machine learning solutions for managing risk online and in person. For risks like fraud and money laundering, Feedzai assesses and detects suspicious patterns in transaction and event data.

Industry impact: According to a Feedzai case study, the company’s technology enabled a major U.K. bank to lower fraud losses by more than $20 million.


fintech studios machine learning applications finance
Image: Fintech Studios

Fintech Studios

Location: New York, New York

What it does: Fintech Studios is an intelligent search and analytics platform providing search for financial professionals across millions of financial and business resources. From blogs and news to research and big data analytics, the platform uses artificial intelligence and machine learning to identify the most relevant information across 49 languages.

Industry impact: Fintech Studios’ variety of products allow financial professionals — from brokers and financial advisors to those at hedge funds and private equity firms — to quickly gain access to the information they need.

More on FintechHow to Catalyze Diversity in Fintech


kabbage machine learning applications finance
Image: Kabbage


Location: Atlanta, Georgia

What it does: Kabbage provides lines of credit to small businesses and has worked with more than 150,000 companies. Its simple application process for businesses (accessible online or through a mobile app) uses machine learning algorithms to determine whether or not an applicant is approved, reducing the possibility for human error.

Industry impact: Since being acquired by American Express in 2020, Kabbage has announced flexible lines of credit for qualifying small businesses up to $150,000, backed by AmEx.


riskified machine learning applications finance
Image: Riskified


Location: New York, New York

What it does: Riskified is a fraud solution for e-commerce enterprises. The machine learning solution identifies bad orders and prevents chargebacks for merchants. The fraud detection solution ensures fewer misidentifications of fraudulent activity and continuously learns new methods of fraud, staying ahead of bad orders and helping businesses retain more customers.

Industry impact: Riskified helped footwear retailer Finish Line to decrease chargebacks by 70 percent, saw more accurate charge declines and ultimately enabled the Finish Line team to focus its time on other business operations and needs.

More on FintechWhat Is Alternative Data and Why Is It Changing Finance?


Square's applications and hardware.
Image: Square


Location: San Francisco, California

What it does: Machine learning performs essential functions even at the most basic level of finance, like point-of-sale retail transactions. Every time you insert your card into a ubiquitous Square stand or terminal, for instance, machine learning and reinforcement learning models operate in the background, flagging sellers to moderate or high levels of risk. The models take in thousands of data points across the ever-building body of transactions to detect new patterns of fraudulent activity.

Industry impact: Square’s dispute management has saved vendors $330 million since 2011, according to the company.


Screens of Varo Bank's iOS application.
Image: Varo Bank

Varo Bank

Location: San Francisco, California

What it does: Neobanks, or fully digital banks, have been active in the United Kingdom for years, but the movement is finally beginning to mature in the United States, thanks to companies like Chime, Current and Varo Bank. Last year, Varo became the first neobank to receive a banking charter in the states, and in 2021, it raised $573 million to broaden its mission of no-minimum-balance, no-monthly-fee banking. Machine learning plays important roles; the company deploys models for fraud detection, recommendation systems and risk assessment.

Industry impact: In recent months, Varo has tripled revenue and doubled its number of customer accounts, to four million, according to the company.

Great Companies Need Great People. That's Where We Come In.

Recruit With Us