Discuss
That TP Lady27 Oct 2026

Human judgement versus artificial intelligence

An AI agent performs a benchmarking analysis and selects 15 comparable companies. A transfer pricing practitioner independently performs the same exercise and selects 11. Only seven companies appear in both sets. Both have applied the same screening criteria and both present technically defensible arguments for their respective selections. Which set of comparables would you use? Would you give greater weight to human judgement simply because it is human? More interestingly, how would a tax authority approach the two analyses?

1 reply

That TP Lady27 Oct 2026

I think this is partially why we are here. We are looking to ensure that there is consistency in how we apply professional judgement when AI can come along, reshuffle established approaches and throw red herrings left and right. In practice, this could help consolidate the internal approach by introducing a third reviewer and building a sustainable checklist: “AI may do this…”, “We choose to do that…”, documenting the rationale behind those decisions. Suppose the two sets of comparable companies produce materially different arm's length ranges and the taxpayer selects the more favourable one. I would be interested in how the selection process was documented, whether the alternative analysis was considered and how the final decision would be defended during a tax enquiry.

Add a reply