According to the House of Commons, there were approximately 5.6 million private sector companies in the United Kingdom on 1 January 2023, i.e. 0.8% more than in 2022. At the same time, number of self-employed workers In the UK, the number of cases has been rising steadily for 20 years (except for a drop in 2020-2022, due to the economic impact of Covid) to reach more than four million.
All these businesses and self-employed people must complete their tax returns for each financial year, which are overseen by the UK tax office HM Revenue & Customs (HMRC).
Unfortunately, HMRC is struggling to respond to enquiries due to staff shortages. This has resulted in long waiting times for calls, averaging 47 minutes.
In George Osborne’s 2015 Budget speech, the former Chancellor of the Exchequer announcement the government’s plan to update taxation, declaring it to be a “revolutionary simplification of tax collection.” This plan later became the Making taxation digital program.
Its goal is to make it easier for individuals and businesses to ensure that their tax returns are accurate and up to date, and the program is currently in effect. expected to be deployed from 2025.
Generative AI and data scientists
To address some of the challenges currently facing HMRC, it has been recruitment of data scientists develop generative artificial intelligence (GenAI) tools that will help tax advisors in their workload.
“We recognise that GenAI has enormous potential and are exploring a range of possible ways to use it, whilst ensuring that we manage the risks to public trust associated with new and rapidly evolving technologies,” an HMRC spokesperson said.
“Where the use of AI may impact our customers, we ensure that the outcome is explainable, involves a human, and complies with data protection, security and ethics standards.”
Generative AI could be used to help HMRC in two ways. The first is to automate tax advisors’ simpler tasks, such as basic data entry, so they can focus their time and energy on the more challenging aspects of their role.
We recognise that GenAI has enormous potential and are exploring a range of possible ways to use it, whilst ensuring that we manage the risks to public trust associated with new and rapidly evolving technologies.
HMRC Spokesperson
“HMRC simply don’t have the people to answer the phone. No one is saying they’re slacking off, but they just don’t have the resources to do it,” says Chris Thorpe, a chartered tax adviser and technical agent for the Accredited Institute of Taxation.
“If you rely more and more on automation, you’re going to need people to pick up the phone, because most of the time the software isn’t going to work or people aren’t going to be able to access it, and they’re just going to need to talk to someone. By relying on automation, you have to have the support at the same time, and that’s unfortunately where people just haven’t been able to keep up.”
AI could also be used to review self-assessment tax returns as they are submitted and look for anomalies. These discrepancies could be simple errors, but they could also be signs of potentially fraudulent practices. Anomalies can arise for many reasons, as real life can be inherently chaotic and unexpected events can occur. As a result, an individual or business’s finances can be impacted by a number of external factors that cannot be predicted, and therefore appear as an anomaly in their tax returns.
HMRC has historical archives of all UK tax records, which it could use to train AI systems. However, many of these records are stored in legacy systems, meaning the hard data needed to train AI would be difficult to extract and format in a usable way.
“The old systems don’t work well together,” says Alison Kerrey, partner at Moore Kingston Smith and chair of the Chartered Institute of Taxation and Association of Tax Technicians Digitization and Agent Services Committee. “The data may be accurate, but it may not be in the right place or going to the right person for you to use the technology effectively.”
The complex challenges facing GenAI
Tax law is also extremely complex. There are basic laws that serve as a foundation, but there is also a wealth of case law that can be subject to interpretation. The situation is made worse by the fact that tax law is constantly evolving.
With each Budget announcement and Autumn Statement from the Government, the UK’s approach to taxation is changed to meet government policy. As a result, what was acceptable a few years ago may no longer be sustainable.
“There’s a lot of legislation, but there’s also a lot of interpretation,” Kerrey says. “The rules will get you a long way in some cases, but then you have to add maybe 40 years of case law and different interpretations, and the balancing of things, which is not very easy to automate. It’s quite difficult to build tools to handle that.”
GenAI is not recommended for tax law advice, as it can sometimes provide convincing answers that are factually incorrect. If generative AI is unable to find an answer, it may combine similar historical examples to hallucinate an answer that appears correct but has no factual basis.
“A lawyer used ChatGPT to produce a series of court records. It was found that the software had completely fabricated them and the judge threw out the entire case,” Thorpe says. “It’s a pretty extreme example, but it shows that practitioners may not have looked at the situation properly and assumed all the numbers were accurate.”
AI will be better than anything we’ve ever seen, but ultimately it will just be a tool that needs to be managed properly. There will need to be some regulation to hold practitioners accountable and to ensure they use AI responsibly.
Chris Thorpe, Chartered Institute of Taxation
The unpredictable nature of GenAI means that human oversight will always be required to ensure that the information provided is fair and accurate. While this is not generative AI, the lack of AI oversight Post Office Horizon Computer Softwarecombined with insufficient critical analysis of the data provided, led to hundreds of sub-postmasters being accused of theft, fraud and false accounting.
“We have already seen what is happening, with all the Horizon and the Post Office“When you rely slavishly on computers,” Thorpe says. “As long as people accept that mistakes will be made on both sides and that those mistakes can be corrected without negatively impacting people, I don’t think there will be a problem.”
Additionally, generative AI works best when it uses hard, absolute data, but it struggles to handle the nuances that humans can bring to an equation. For example, a person might be self-employed or an employee, but they might also be a part-time employee who also owns a small business—cases like this can make it difficult for GenAI to rationalize the situation.
Another problem is that some businesses are not digitally connected. They may operate in remote locations where internet is not available, or simply do not want to use digital tools. For example, a timber merchant near Derby uses a slate and chalk to calculate payments and only accepts cash.
Automate tasks, but with human supervision
The bottom line is that the judicious use of generative AI can reduce the workload by automating simple tasks. This will allow tax advisors to focus on the complex elements of tax law and edge cases that AI would struggle to handle.
“AI will be better than anything we’ve ever seen before, but ultimately it will just be a tool that needs to be managed properly,” Thorpe says. “There will need to be regulation to hold practitioners accountable and to make sure they use AI responsibly.”
Ultimately, HMRC does not intend to replace tax advisors with generative AI. People with experience and knowledge of UK tax laws will still be needed to respond to queries, whether by phone, email or instant messaging. GenAI can help meet the demands placed on HMRC by automating simpler tasks and providing basic anomaly detection, but a human element will still be needed to maintain oversight.
“AI can miss certain complexities and sometimes return wrong results,” Kerrey says. “Any future use of AI will need to be able to get around this hallucination.”
