For Intuit, data science is a bread-and-butter technology. By implementing artificial intelligence and machine learning within our applications, we’re giving customers new ways to get control over their financial lives, from getting the full tax refunds to which they’re entitled, to getting quick access to capital to supplement small business cash flow. Intuit Chief Data Officer Ashok Srivastava recently spoke with long-time ZDNet contributor Tonya Hall about how Intuit thinks about data science and how we put it to work for our customers. You can find the full video here; the following Q&A excerpts are adapted from the conversation.
What are some of the ways Intuit customers are currently benefitting from AI and ML?
Ashok: One area we’re focusing on is developing new technologies to help people with personal finance tasks such as accounting and tax preparation. Many people don’t realize that the U.S. tax code is a full 80 pages long—much more than an ordinary person could be expected to master. We’ve built a capability to convert the tax code to computer code so that TurboTax can apply the rules for them. This makes it possible for people to complete their tax returns quickly, efficiently, and accurately without having to become experts. For people who are self-employed, from Uber and Lyft drivers to professional services providers, the ability to automatically flag allowed deductions can yield thousands of dollars in savings each year. Implementations like these are pervasive across our entire portfolio, including TurboTax, Mint, QuickBooks, and other products.
How does AI/ML make loans available to people who haven’t been able to get them in the past?
Ashok: Half of small businesses fail in the first five years. These are the foundation and fabric of our economy—it’s critical to give them the help they need to grow and prosper. We built AI/ML technologies to take data from QuickBooks, match it with third-party data, and use it to make more nuanced and timely underwriting decisions. Where a traditional bank might say no, we can use our insight to quickly provide credit to people in need, like a small business owner who needs a short-term loan to make payroll at the end of the week. That’s one of the best uses of data science, addressing a fundamental need people have in the small business world.
What role does transparency play here? Will people accept AI/ML outcomes if they can’t audit the algorithm?
Ashok: It’s very important to be transparent—that’s why we’ve built techniques that are completely explainable. I wouldn’t want a machine to make decisions for me without knowing what’s happening. That might not matter as much in other areas; you don’t necessarily need to know how an image ID algorithm is telling dogs from cats. But in finance, we really need to spell things out.
As we look at the ways humans and AI interact, do you anticipate breakthroughs on the horizon or incremental changes?
Ashok: Today, we see three broad models for these interactions. One is human-centric, where a person in charge is getting some amount of advice or assistance from a machine. Augmented intelligence is another more exciting model, where the human is in still charge, but gets rich, relevant, and specific information in real time to help them do a better job. This is a vast area, not only in the applications Intuit is delivering today, but also in areas from aviation and space programs to medicine. The third area is autonomous AI, where machines make decisions with human authority behind them. Each of these models has a valuable role to play in various use cases, and I do anticipate important breakthrough technologies in all three areas.
To view the full interview, including how Intuit has expanded its data science capability over the past year, you can find the video here.