Share this content

MTD - the End Game?

The potential use of AI on MTD data

Didn't find your answer?

It's Friday, so time for a rant.

I see that Apple is in the news because their iPhones will now scan your data for evidence of child abuse. Naturally that is to be welcomed. Isn't it? 

I wonder what is the true intention of Apple's AI processes - is the child abuse angle a PR spin so that nobody can object? Would we all be welcoming this if they had said that your iPhone will be scanned to ensure that you have paid your council tax, or check whether you drink more than the recommended government amount, or if you have been socialising while unvaccinated?

I find this slightly troubling because first of all, who appointed Apple as the guardian of our morals? And secondly, how long will it be before governments in various jurisdictions are forcing Apple to hand over this data? Or perhaps Apple will willingly undertake this process for a fee - raking through our data to supply governments with our information.

Now personally I don't have a stake in this because I don't use a mobile phone. But where Apple leads most software developers will follow and I can't help but wonder how cloud accounting software will develop the use of AI. Can we be sure that client data in the cloud is really protected against future abuse?

A recent article on this site has criticised Xero for adding features that nobody wanted. I am not singling out Xero here but we have no power to stop any software provider from introducing AI analysis (offered as a client benefit), using it for their own analytical purposes and passing the results to anyone willing to pay the fee. If you think we will have a choice in this, think again. 

In fact, this could actually provide a rationale for MTD. We have all been asking what is the point of MTD - maybe it's now becoming clear. Chatham House, a respected international policy think tank, has a short blog article suggesting we are sleepwalking into an AI moral maze. I think they may be right.

Replies (6)

Please login or register to join the discussion.

By SXGuy
06th Aug 2021 07:42

I don't see any AI ever being able to work out exactly what something was unless it's a regular expense.

How would it know whether to capitalise something? It won't.

I think with the introduction of things like auto tagging bank entries, that imo is the closest to AI we should try to get.

Thanks (0)
06th Aug 2021 08:25

We may be there already.

We have had VAT inspections for a while that consist of sending HMRC a backup from your accounting system and HMRC's software looks at it and comes back with questions about specific entries. Several years ago, pre MTD, one of our clients had a VAT visit from an inspector who had been involved in developing the software to do this, even then they were aiming to obtain the maximum amount of information from the accounting records and not all of it was VAT related.

While AI won't necessarily be able to pick out the correct accounting treatment for specific transactions, it can identify areas for further investigation - a lot of the VAT inspection was looking at ratios within the records and it wouldn't be too hard to pick out payments to a particular business or individual.

MTD and the compulsory computerisation of records just makes the process easier for HMRC

Thanks (0)
By Tornado
06th Aug 2021 11:58

I am intrigued by the way that the term AI (Artificial Intelligence) is being used regularly when the software being referred to is simply advanced computer analytical programs.

AI is where I can hold a meaningful conversation with a computer that understands human beings implicitly and has emotions and can even understand our idiosyncrasies such as not running a business as a text book would suggest, or doing things that are illogical.

The problem with leaving it all to the computers is that human beings step back and let the computers do all the work, even is the work turns out to be utter rubbish. The real danger, of course, is that eventually these super analytical programs could rule our lives indiscriminately if they are left to themselves. That is the danger, not AI.

Thanks (0)
By Mr_awol
09th Aug 2021 09:06

AI analysis (and sale of anonymised results dressed up as benchmarking data) is almost certainly somethign the likes of Xero will be doing - if they dont already. Why do you think NatWest wants to buy Freeagent? To offer it as a perk to customers?

Our data is important and valuable and cloud companies will milk it whether we like it or not, as soon as they feel their user base is locked in enough. No different to Strava recently removing certain features previously available on the free platform, including some leaderboard result data. They are happy to try to force people to pay for their (expensive for what it is in most cases) premium offering but still happy to harvest data from the very users they are now cutting out of the experience. Cloud software providers will be doing the same.

Thanks (0)
By johnhemming
09th Aug 2021 12:40

Automatic analysis is possible for some transactions, but not for all of them.

Thanks (0)
By Catherine-Xero
13th Aug 2021 15:45

Wanted to pop into this thread and share this recent post on our blog about our commitment to responsible data use - might shed some light on how our team is approaching this > Feel free to post any direct questions about this on the blog so that one of team who knows more than I do can answer them :)

Thanks (0)
Share this content