Digital labour: AI and robitics | KPMG | UK

Digital labour: AI and robotics

Digital labour: AI and robitics

Cognitive automation and artificial intelligence has the potential to completely transform the tax world. Find out how companies are getting to grips with the new technology.

1000

Director, Tech Solutions, UK Head of Data Engineering

KPMG in the UK

Contact

Also on KPMG.com

digital labour ai and robotoics

Finance functions are marching to a new techno beat. Many companies are reaping the benefits of automating routine work through robotic process automation, or RPA. Reducing cost and improving speed – what more could a company ask for?

Quite a bit, it turns out. Having got wise to the benefits of RPA, a number of companies are turning to the next step of the automation journey: cognitive learning. This uses robotics and AI techniques to simulate human cognitive processes, bringing statistical models to play to train machines to perform these same processes based on real decisions that humans have made.  

This is where the real value sits. Cognitive is here and companies are already getting to grips with it. KPMG has set up projects in which machines are trained to take over decision-making, with humans staying in the loop mainly to provide quality assurance over the decisions the machines have arrived at. 

The best use for cognitive automation is harvesting rich training data sets. These are sets that represent decisions that humans have made as well as the decision outcomes determined from the data that they have used to make the decisions. 

A classic example in the commercial sector is assessing insurance claims. Much of the data comes from a structured system but there is also a lot of unstructured data (such as photos and accident reports). Machine learning takes all the data, breaks it up into features, and looks for patterns in order to arrive at a decision. 

Tax functions are ideal for cognitive. At the simpler end of the scale, transfer pricing could use bots to collect and analyse data rather than sending out questionnaires or having meetings or calls. Over time the bots would learn what questions to ask and how best to interrogate data. And of course, they could run 24 hours a day, seven days a week.

At the more sophisticated end, watch out for AI tax advisers. These would start by ingesting the whole corpus of law, cases and regulations. Every time a client asks a question the output would be included in the corpus in order to statistically train it to produce the right answer. When new questions are asked, machine learning could be applied to understand the intent of the question, then to find out when it had been asked before and where there has been a positive result to the question. 

This might sound like science fiction but sophisticated AI like this is already in place at legal firms and is very close indeed for tax.

This technology will cause a revolution in the tax world. Investing in it will allow firms to adapt to the new needs of the market. It will attract a new diverse tax resource. And this will not be optional: it won’t be long before robotics becomes the only tax tool in town: the next generation of tax advisers simply won’t know another way of working.

Tax authorities have started to use AI to analyse tax returns and vast quantities of data, so companies and tax advisors should be doing the same. How else will companies  understand what the tax authorities may interpret and require justification and support? Consider for instance the analytics that could be produced from submitted country by country reporting returns.

This future is close, if not already upon us. Tax advisers are seriously investing in robotics capabilities; tax functions need to start thinking about the implications of this now. They need to consider how they are managing taxes in their global business and what their priorities are. If they have not already invested in RPA they should look into this,  seeing what is available and assessing the benefits it can bring. Then keep an eye on what is happening in the market, what competitors are up to – and what tax authorities are investing in. 

A final suggestion: companies should make sure to check what the global organisation is doing. The wider finance function with its access to bigger budgets might already be investing in automation and analytics. Is there any way that tax could piggy-back off this investment? This would make the tax section part of the group’s global strategy and toolkit, accessing greater value for the group as a whole. 

Tax Matters Strategies - Future of Tax

Tax Matters Strategies - Future of Tax

Change is coming. Are you ready to rethink your tax function?

Connect with us

 

Request for proposal

 

Submit