There could also be a number of challenges that tax authorities in India may face when using artificial intelligence. These can include the need to balance centralisation of tax assessment with on-ground experience, uncertainties around being able to develop models which deliver a positive impact on performance, the varying range of choices in AI solutions, and the need for skills to leverage those advanced solutions.
India may soon become the first country to use artificial intelligence and machine learning in the tax assessment process. Finance Minister Nirmala Sitharaman has announced that the government will deploy a faceless assessment system based on AI and ML, starting October 2019. The overall process will increase the accuracy and transparency of India’s tax assessment process, thereby improving the tax base and compliance.
As per the announcement, data within income tax returns, statement of financial transactions and from other sources will be analysed to look for anomalies so that tax compliance is achieved. All of these would require the tax department to process data generated from billions of financial transactions taking place every day in India. The adoption of AI is going to get rid of the existing inefficiencies in the tax processes as well as create insights that can strengthen tax collection and prevent tax evasion and fraud. For AI to succeed, it needs data and clearly, with one of the largest tax-paying population in the world, the government would have all the data it needs.
Faceless Electronic Portal To Scrutinise Tax Assessment
If any anomaly is found in a tax payer’s assessment, there will be automatic scrutiny done using faceless electronic portal wherein a structured questionnaire based on the anomaly will be sent to the taxpayer automatically. Powered by natural language processing capability, the faceless portal would take in the answers from taxpayers. If the system is satisfied with the answers, the case will be closed, otherwise, it will be assigned to an income tax officer for a further enquiry.
It is to be noted that the income tax department (ITD) had earlier identified many citizens who had not paid their tax liabilities when filing their tax returns for the assessment year 2018-19 using data analytics. This was part of the non-filers monitoring system (NMS) deployed by ITD, which monitors individuals for high-value transactions and potential tax liabilities using data analytics.
AI/ML Based Tax Assessment Is A Logical Step To Government’s Recent Policies
Like most countries, India’s Income Tax Department has been dependent on human tax assessment officers to assess tax returns filed by individuals, which leaves scope for inefficiencies and tax evasion on a large scale, all of which have magnified the compliance workload for businesses and the tax department. India’s existing tax environment requires increased transparency across different departments and tax authorities, and using AI-led automation can be tremendously helpful.
The AI/ML tax assessment system is a logical next step to policy decisions implemented by the Indian government in recent years. This includes full data-localisation by foreign companies conducting business in India, digitisation of tax filing systems, and the government’s focus on deploying electronic payment systems and making large cash transactions illegal.
To roll out the use of AI/ML in taxation, the government is working to integrate data from Ministry Of Corporate Affairs, Central Board of Direct Taxes (CBDT) and the various systems therein within a year, particularly before next year’s budget. The direct tax panel, headed by Akhilesh Ranjan has recently also proposed to introduce collaborative compliance in the direct tax administration, which would integrate data from banks, financial institutions and the goods and services tax (GST) network to ensure that the scope of taxable income increases.
Challenges That Government May Face In Deploying AI For Tax Compliance
There could also be a number of challenges that tax authorities may face when using artificial intelligence. These can include the need to balance centralisation of tax assessment with on-ground experience, uncertainties around being able to develop models which deliver a positive impact on performance, the varying range of choices in AI solutions, and the need for skills to leverage those advanced solutions. For this reason, according to a media report, sources within the tax department fear that the hasty deployment of AI and ML may lead to a situation similar to the havoc caused by the poor implementation of GST.
Others are concerned that the tax officers may become complacent and will not be keen on hearing the assesses’ viewpoint but rather just depend on the results and data crunched by AI models, which may undermine the whole process. Since AI is inherently probabilistic, there may inevitably be cases where the machine learning model comes up with a finding that could deem to be wrong, and so human supervision will still be important. For the income tax department, the starting point will be to have a precise data strategy to suit the compliance requirements along with training that will need to be tailored to the taxation policy and taxation law in the country.