Article: AI — The Sleepless, Relentless New Force in Taxes

Photo Source: Pexels

This week, the US is wrapping up its tax season. Culminating in April, every citizen files their tax return forms with the Internal Revenue Service (IRS) for the previous year. Tax returns must include wages, profits, and other forms of income as well as losses from investments, businesses, and charitable donations. These tax returns inform the government of how much tax it can take based on a large set of rules enacted by Congress. The US tax code, in fact, comprises a highly complex system. The current federal tax code Public Law 117-154 fills 6,871 pages, and with the official tax guidance and regulations included, the code balloons up to 74,000 pages. The code and guidance still does not include state and local tax laws. (irisreading.com)

The IRS has an army of tax auditors and artificial intelligence tasked with finding those who did not pay everything they owed. The law punishes underpaying taxes with fines, repayment obligations with interest, and even jail time. The US Treasury Department collects $4.9 trillion in taxes annually, but the government still spends more than it takes in from taxes and other income sources. While the IRS does have plenty of tools to catch tax evaders, they only have so many resources to go through every tax return. (fiscaldata.treasury.gov) To ramp up tax revenue, Congress recently passed a law titled the Inflation Reduction Act, which includes $80 billion to be given to the IRS over the next ten years in addition to their $12.6 billion yearly stipend as of 2022. (cnn.com) Over half of the $80 billion will go to greater enforcement, such as more auditors and officers for collecting taxes owed and developing new technologies to assist the IRS.

The IRS already uses artificial intelligence in its current operations. In 2018, the IRS began a ten-year, $99 million contract with the data and artificial intelligence company Palantir. Palantir, headquartered in Denver, Colorado, initially helped the US Intelligence Community and Department of Defense in counter-terrorism activities. Now, it also uses large data sets and advanced analytics to detect tax fraud. Palantir, combined with the IRS, accesses the government's vast data on its citizens and uses this information to find suspicious activity. According to the IRS, "Contact Information (name, address, phone numbers, Date of Birth (DOB), and IP addresses) are used to identify individuals directly and indirectly related to the investigation." (irs.gov) Suspicious activity and criminal records flag a person for a potential audit. The IRS also works with IBM to use AI to scan paper tax returns using machine vision, significantly reducing the time it takes for IRS workers to transcribe paper filings. According to IBM, their machine vision and character recognition system processed 140,000 paper returns without human intervention. (ibm.com)

The US government collects taxes each year, and the process for most people culminates in April, where they file their personal and business tax forms and reconcile any missing tax payments. The hundreds of millions of tax filings generate over $4 trillion in revenue each year. The revenue still does not cover what the government spends. To make sure no cent is left uncollected, the IRS attempts to find those who wittingly or unwittingly evaded paying their legally required amount of tax based on the US Tax Code. In addition to traditional methods of detecting tax fraud, the IRS has embraced artificial intelligence to enhance its ability to find those not paying their share. Apart from using tools such as IBM's machine vision for increasing efficiency, the IRS entered into a contract with the big data analytics company Palantir to use artificial intelligence to sift through the billions of records and banking activities of US citizens to detect patterns of fraud. The system promised to help find tax evasion at the highest level; however, low-income individuals and families still receive the bulk of audits from the IRS. According to the University of Syracuse Taxtrac system, low-income wage earners have a 5.5 times higher chance of an audit by the IRS than all the rest of the taxpayers combined. (trac.syr.edu) AI may impact the auditing process today, but it may not be ready yet for the complex taxes of higher-income earners.

Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high-level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.

You can buy his book on Amazon in paperback and in kindle format here.