Skip to main content [aditude-amp id="stickyleaderboard" targeting='{"env":"staging","page_type":"article","post_id":2161223,"post_type":"guest","post_chan":"none","tags":null,"ai":true,"category":"none","all_categories":"ai,bots,","session":"B"}']
Guest

How this chatbot powered by machine learning can help with your taxes

Image Credit: Shutterstock

Tax season is around the corner, and for most Americans, it involves dealing with the complex tax code, an accountant, and maybe friends who claim to be tax experts. According to IRS statistics, there were 507 million visitors to the irs.gov website in 2016, a 3 percent increase when compared to 2015. Furthermore, Americans spend 6.1 billion hours and $233.8 billion complying with the tax code. With complexity comes confusion and frustration, leading many taxpayers to turn to tax preparers or tax preparation software. Those who don’t are at a disadvantage because most Americans are unaware of which deductions to consider, dependents to claims, student loan amounts to deduct, or where to file.

Considering the advances in technology and promised tax policy reform from the Trump administration, consumers expect a simple tax software solution. What will this tax solution look like? Will technology be able to reduce complexities for individuals and businesses? Will advances in artificial intelligence play a part? What role will a chatbot play?

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":2161223,"post_type":"guest","post_chan":"none","tags":null,"ai":true,"category":"none","all_categories":"ai,bots,","session":"B"}']

Machine learning is a branch of artificial intelligence that gives computers self-learning ability through pattern recognition and periodic human intervention. The idea behind machine learning is constructing computer algorithms that automatically improve themselves by finding patterns in existing data without explicit instructions. Machine learning relies entirely on data. As the quantity and quality of data increases, so does the machine’s predictive accuracy.

If a machine learning model were to study tax and financial inquiries for 10,000 or so taxpayers, it could learn what deductions to take, how those issues affect income tax liability or refund, and offer guidance. While tax rules are complicated, there are well-defined IRS publications and use cases. Furthermore, the majority of the things that affect the average taxpayer’s return are already digital and available online — including paystubs from payroll services, investment gains and losses from brokerage accounts, and mortgage payments — and the information can be included in an algorithm. One can:

AI Weekly

The must-read newsletter for AI and Big Data industry written by Khari Johnson, Kyle Wiggers, and Seth Colaner.

Included with VentureBeat Insider and VentureBeat VIP memberships.

  • Refine the machine learning algorithm when there is a change in accounting/tax laws
  • Apply the algorithm to a specific industry or company (every industry and company is different)
  • Review the output of the algorithm and edit them based on judgment.

With the advent and application of machine learning, accountants will no longer need to classify transactions (e.g., selecting the right account codes or determining the tax deductibility of an item). A machine-learning model can guide a taxpayer and improve assessments for subsequent taxpayers, thus reducing the role of a tax adviser and possibly replacing tax preparers.

In the coming years, there will be a “machine” that has access to a large volume of tax data, tax codes, tax litigation, and relevant facts for a business enterprise. With this data, it will be possible to build industry-specific tax software. Upcoming tax law changes and reforms would have already been analyzed and their impact explained and accounted for.

Because the tax code is dense, a machine learning solution will be ideal to solve the pain point for taxpayers. For example, a machine learning solution can handle:

  • Question answering. Given a human-language question, determine its answer. Typical questions have a specific right answer (such as “I drive for Uber — is my wait time tax deductible?”).
  • Relationship extraction. Given a chunk of text from the IRS code, identify the relationships among case law and FAQs.

Because of the complexity of the tax code, my team and I built a rule-based machine learning chatbot named AskMyUncleSam to answer tax and financial queries to make life easier for the 212 million American tax filers. The tax code is full of incentives that taxpayers can use to their advantage to pay less in taxes, but taxpayers are usually unaware of these advantages. Getting a college education, home ownership, and paying for childcare are just some of the many activities that are tax advantages for individuals. For businesses, the tax code is even more generous in terms of allowable business deductions.

With AskMyUncleSam, you will know what tax benefits are available to you and what actions you could take to reduce your tax bill by asking Sam a tax or financial related question. Our goal is to build up a database of 8,000- 10,000 tax-related queries to serve the interests of taxpayers throughout the entire year, and present this analysis and advice to them in a timely and easily understood way. Filing taxes shouldn’t be a confusing and stress-inducing process, and we at AMUS aim to reduce the opacity of the tax code and allow easier tax filings for millions.

Technology can’t solve every part of what makes taxes so complex and painful. In the coming years we will see innovation from technology-savvy CPAs and accountants, as well as software titans such as H&R Block, Intuit (TurboTax), Thomson Reuters, and TaxSlayer, to address the taxation pain points.

[aditude-amp id="medium1" targeting='{"env":"staging","page_type":"article","post_id":2161223,"post_type":"guest","post_chan":"none","tags":null,"ai":true,"category":"none","all_categories":"ai,bots,","session":"B"}']

VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn More