Artificial Intelligence and Automation at the IRS

Co-Author: Steve Deering, IBM Associate Partner, Data and Technology Transformation

Leaders recognize that technology is critical to improving taxpayer service, closing tax gaps, and addressing workforce challenges. The most recent piece in the series focused on opportunities around AI and Automation in Tax. This one takes a look at how the Internal Revenue Service (IRS) took advantage of these capabilities to deliver a more seamless taxpayer experience.

The delivery of public services by the government continues to evolve as citizens increasingly look for more personalized and seamless experiences. The last three years were a tailwind, pushing government to rethink service delivery models. The IRS touches every household and is no different. The IRS strategic goals are to make the IRS more accessible, efficient and effective. With help from increased funding, the IRS is adopting industry leading technology and increasing the efficiency and currency of technology investments.

In 2020, the IRS faced significant paper tax return processing backlogs. Tractor trailers full of paper were sitting at processing centers to be opened, scanned, sorted, properly inventoried, and manually processed onsite. Tax return processing delays slowed taxpayer refunds by as much as six months or longer. And although the increase in electronic filing of tax returns has greatly reduced the number of paper returns, the IRS receives, paper tax return volume is still significant. The IRS established a pilot program to automate the scanning, validation and processing of its tax year 2020 and 2021 paper tax returns.

Digitization went well beyond paper scanning and the automated extraction of data from paper tax returns. The solution processed nearly 140,000 paper tax returns at a significantly higher rate of quality than today’s human transcription – providing the IRS with the simplicity and efficiency needed to tackle its backlog challenge. For 2021, tax forms processed during the pilot "Modernized e-File (MeF) system" accepted 76 percent of paper tax returns processed without human intervention. Technologies used for the project also laid the foundation for the possibility of future anomaly and fraud detection during the tax return intake process – even for paper tax returns.

More importantly, the IRS demonstrated how new efficiencies can deliver faster taxpayer refunds – improving taxpayer experience. This was accomplished using optical character recognition technology and AI to extract data from the digital images. With data extracted, a series of business rules, automated validation sequences, and compliance routines prepared data for ingestion or further review by humans for handling returns with errors.

In today’s rapidly changing, technology-enabled world, citizens have become used to services that are only a click away. Tax agencies, like others in government, are under increasing pressure to keep pace. Digital transformation is a marathon, not a sprint. It’s steered by continuous technology innovation that presents new, more effective ways to conduct business and deliver services. Complex challenges and growing citizen expectations make it imperative that agencies maintain the accelerated pace of innovation to deliver value today and tomorrow.

Read the next blog in our series.

Image by xb100 on Freepik