• Intuit Blog
  • All Posts
  • News
  • Using Artificial Intelligence to Lessen Work and Drudgery: How We Apply Data Science to Tax Preparation

Using Artificial Intelligence to Lessen Work and Drudgery: How We Apply Data Science to Tax Preparation

Each year, more than 253 million people in the United States will file their Federal income taxes.1 Many of us anticipate the prospect of this activity with significant anxiety. Taxes are a complex endeavor, and worries about appropriate compliance can generate significant stress for tax filers.2   Because of these challenges, it seems like a

Each year, more than 253 million people in the United States will file their Federal income taxes.1 Many of us anticipate the prospect of this activity with significant anxiety. Taxes are a complex endeavor, and worries about appropriate compliance can generate significant stress for tax filers.2

 

Because of these challenges, it seems like a good idea to use data science and automation to help alleviate the work involved in income tax filing. After all, shouldn’t we use this technology to take care of less-fun activities (like taxes), instead of automating things that are fun (such as driving automobiles)?

 

In my last post, I mentioned how we are using artificial intelligence and machine learning to make tax preparation less painful. This time, I’d like to explain how we’ve done that in a little more detail.

 

In 2007, Intuit used artificial intelligence and computational linguistics to help consumer’s answer common tax questions. Through the application of natural language processing and natural language understanding, we would match our customers’ questions with answers provided by various tax experts.

 

The next step, which we started in 2011, was the creation of a tax knowledge engine (TKE) which encodes large amounts of conditionally-related tax constraints into tax knowledge graphs. The engine can then intelligently handle the many different types of tax situations faced by consumers.

 

Similar to how human experts make tax decisions, TKE executes a collection of algorithms over the graphs to address three problems for any given tax situation:

  1. What’s missing (complete data)
  2. What’s wrong (contradictions)
  3. Why (explaining recommendations)

 

By intelligently determining what questions to ask the tax filer, TKE minimizes the number of questions needed to get the information needed and make the appropriate recommendations. This functionality also lets TurboTax explain back the computational results in a fully personalized manner.

 

For example, through this technology, we can help our customers understand why they qualify for the earned income tax credit or why they are punished for a late payment… and all sorts of other factors or decisions that make taxes confusing to most of us. By building the engine to power this process and the various factors, we hope to increase each person’s confidence in the tax filing process and help them understand why certain factors may have changed for them from year to year.

 

As data scientists, we have the opportunity to significantly reduce the anxiety for millions of customers every year, and we are looking for talented data technologists who are willing to build the next solution that can help their friends, family and community. If your expertise fits this area and you are interested in making a real difference in the financial lives of others, I encourage you to check out our career site to learn more.

 

1 Fiscal year return projections for the United States: 2017–2024, Fall 2017, IRS Statistics of Income Division within the Research, Applied Analytics, and Statistics organization. https://www.irs.gov/pub/irs-soi/p6292.pdf

2 Tax compliance and psychic costs: behavioral experimental evidence using a physiological marker, February 2016 Journal of Public Economics. https://doi.org/10.1016/j.jpubeco.2015.12.007

 

 

Ashok Srivastava

Ashok N. Srivastava, Ph.D., is Senior Vice President and Chief Data Officer at Intuit. He is responsible for setting the vision and direction for large-scale machine learning and artificial intelligence across the enterprise, to help power prosperity across the world.