Demanding the right amount – are your numbers hallucinating?
In this issue
- Getting calculations right is key when making demands or claims under contracts
- Failure to do so may have serious repercussions
- But whilst AI tools can perform calculations quickly, do they always get the numbers right?
About the author
Stephan Smoktunowicz is a banking & finance partner at the London office of the international law firm, gunnercooke. He helps businesses & people in the UK and across the world to achieve their commercial objectives and safely navigate issues that arise throughout any transaction lifecycle. He also provides training and mentoring to businesses, professionals and students on related topics.
Introduction
Used appropriately, AI tools can create business efficiencies. They can also give the impression that they really know their stuff.
But where you need to demand or claim monies under a contract, getting the amount correct is important to avoid the risk of that demand or claim being ineffective.
Whilst much of the focus of using AI tools can be on checking that ‘words’ are right, as explored in this article, checking the ‘numbers’ is equally as important.
Why the numbers need to be right
Monies payable under contracts come in all shapes and sizes, for example:
- fixed sums payable on set dates
- interest
- default interest
- fees
- payment uplifts/performance payments dependent on criteria being met (e.g. earn outs under sale and purchase agreements)
- early exit fees
- costs
- expenses
- indemnity amounts
Where a payment is missed or there is a right to accelerate payment under a contract, a party might seek payment via raising a formal demand or court claim for breach of contract.
However, if numbers have been incorrectly calculated, this could provide a legitimate reason for the payee to dispute, delay, or challenge the demand or claim.
Whilst it may be possible to overcome this in some circumstances, not only does this result in more cost and time in the recovery process, in a worst-case scenario, getting the calculation wrong could prejudice the ability to recover anything at all.
Why AI might give the impression it has got the numbers right
AI tools generally provide answers to questions in plain English. So, if you ask an AI tool to analyse a calculation based on what a contract says and ask it to breakdown the calculation for you, the chances are that you will be given a full calculation and breakdown, including, for example, the amount of accrued interest on a principal amount payable and the dates under which that interest accrued.
On the face of it, the calculation breakdown may look hugely impressive. The AI tool might also cross-reference exactly how it came to its conclusions, which could then be sense checked against the contract for accuracy.
The crucial extra sense check that might get missed
However, and of utmost importance, even if the AI tool has identified all of the correct payment types (e.g. principal, accrued interest and early exit fees), there is no guarantee that the actual numbers and overall calculation presented by the AI tool will have been correctly calculated. For example:
- has any calculation for accrued interest due over a period been double checked against the contract and contractual rate(s)?
- if rounding up or rounding down methodologies are provided in the contract, have they been included in the calculation correctly?
- even if sub-components are correctly calculated, has the overall figure been added up correctly?
Some practical tips
No matter how much you rely on AI tools to assist with workloads, calculations used in formal legal documents still need checks and balances around them. The following considerations may help to ensure that your numbers are accurate:
- Only treat AI generated calculations as drafts. Always check the underlying contract and use calculators/spreadsheets or other precise calculation tools to check the maths
- Always consider the type of document you are inserting a calculation into – for example, accrued interest calculations in a demand under a contract compared to calculations in statutory documents such as proofs of debt may require calculations to be made up to different dates (e.g. date of a company’s administration rather than date of demand)
- For complex calculations, consider creating and saving a clear itemised calculations table or schedule, which clearly sets out the sub-components of the overall calculation and basis for it
- Consider adopting a two person check for high value/complex calculations
- Before signing off any calculation – stop – think – is this the true amount of the demand/claim? You may have to certify that it is in court documents
- Consider whether you should be using AI for the calculation exercise in any event – e.g. are there contractual/other restrictions which prevent you from doing this?
Conclusions
When used appropriately, AI tools can reduce the time needed to prepare demands and claim documents.
However, never assume that an AI tool has got your maths right, even if you ask the AI tool to double check its calculations.
Whilst AI time savings may create business efficiencies, those efficiencies may ultimately count for nothing if your claim or demand is successfully challenged and you can’t get your money back.
NEED HELP OR ADVICE?
GET IN TOUCH
If you need any help or advice in relation to the matters discussed in this article or any other banking and finance law related matters, please do not hesitate to contact Stephan by email at stephan.smoktunowicz@gunnercooke.com.
You can also find out more about Stephan’s practice and experience here and connect with him on Linkedin here.
Important:
This article is for information purposes and contains personal views only – it does not constitute legal or professional advice, nor may it be read, taken or relied upon as such.
Where this webpage contains links to external sites and/or resources, these links are provided for the purposes of providing readers with additional information only. The author has no control over the content of any of those sites or resources, and accepts no responsibility for them or for any loss or damage that may arise from any person’s or entity’s use of them.
No representation, warranty or guarantee of any kind is given by the author in connection with all or any of the content of this webpage and readers should always seek their own professional advice. By visiting this webpage you agree that the author shall not be held liable for any direct, indirect or consequential loss or damage incurred by any person or entity in connection with this webpage or any external sites and/or resources linked to it.