This raises two compelling questions
1. Is technology now capable of making better decisions than humans?
2. Can it deliver the best possible outcomes in complex tax scenarios?
As Graham Tilbury, Tax Technology Expert explains
“Tax Technologists often talk about AI for Tax when they really mean machine learning, and they invariably over promise and under-deliver
‘Deep Tax’ is the very opposite of this approach. Using specially developed algorithms, it tackles tax problems in a way that a person isn’t capable of replicating, delivering better outcomes, and using a proven technology that far more deserves the description of Artificial Intelligence. Technology can make decisions in tax, but only if you use the correct technology for the problem, and this technology is the right solution.”
Proving the Capability: Using an HMRC Example
To demonstrate what technology can achieve, we analysed an example from the HMRC manuals. This example illustrates the rules for group relief surrenders with non-coterminous APs. While this scenario involves just five companies, with only two sharing the same AP dates, it demands complex, interconnected decision-making regarding surrender sequencing and loss allocation.
Here’s how the outcomes compare between the HMRC approach and Deep Tax processing:
The results were as follows (with the workings shown below).
Analysis of Results
- Decision-Making Capability
Deep Tax successfully automated decisions traditionally made by tax professionals, achieving better optimisation than the HMRC example. - Processing Efficiency
The entire computation required less than one second, contrasting sharply with the manual calculation burden evidenced by the HMRC manual’s five pages of workings. - Optimisation Quality
The £67k remaining profits relate entirely to non-overlapping AP periods, mathematically proving these results cannot be improved upon, as no further losses can be surrendered.
Addressing the Core Questions
- Has Deep Tax made the “best” decisions?
The mathematical impossibility of further optimisation in the non-overlapping periods confirms that it has achieved optimal results - Has it made “better” decisions?
While the numerical improvement over the HMRC example is clear, it’s important to note that the HMRC calculations primarily demonstrate rule application rather than optimisation. However, our experience with similar live data scenarios confirms that this level of optimisation is extremely challenging to achieve manually, even with CT software functionality.
Beyond the Example
While this demonstration focused on single-period scenarios, Deep Tax’s capabilities extend far beyond. The case study referenced on our Tax Technology Fair and Case Study page demonstrates its ability to handle substantially more complex scenarios involving multiple periods and intricate group structures.
Conclusion
The results demonstrate that technology can not only match but enhance human decision-making in complex tax scenarios. Deep Tax has proven its ability to deliver:
- Mathematically optimal results
- Significant efficiency gains
- Consistent performance at scale
These capabilities represent a significant advance in tax technology, moving beyond traditional methods to achieve true intelligent decision-making.
Workings
HMRC (uses daily time apportionment)
Deep Tax (uses daily time apportionment)
