Tax Guru – Ker$tetter Letter

Helping real people win the tax game.

  • Enter your email address to subscribe to this blog and receive notifications of new posts by email.

    Join 668 other followers

  • Blog Stats

    • 292,479 hits
  • Posts By Day

    July 2020
    M T W T F S S
     12345
    6789101112
    13141516171819
    20212223242526
    2728293031  
  • Subscribe

  • Special Pages

Can A.I. Handle Accounting?

Posted by taxguru on July 29, 2020

Over the past few years, as robots and artificial intelligence (AI) have made huge inroads into several industries, I have seen many CPAs and bookkeepers fret over the possibility of having no work to do.  DIY tax return programs and AI were supposedly going to make them obsolete and they sincerely worried about their futures.

Whenever possible, I have assured these doomsayers that as long as they are not just “form-fillers” they will never run out of work.  Tax and accounting rules and regulations will continue to grow more and more complicated until the end of time.  There will thus always be a need for people to apply their knowledge and experience to solving and complying with those rules and regs. 

Having been in this business for almost 45 years, I have seen and been a part of the evolution from doing books manually in big heavy ledgers, to in-house computers,  to working in the cloud.  These have all been means of recording transactions for businesses, much like going from a pencil to a typewriter.  There has always been a need for humans with proper accounting expertise to determine where and when transactions should be recorded in a company’s books.  While it may be possible for a computer to be programmed to play games such as Chess, being able to analyze and properly book accounting entries is too much science fiction for me to accept.

However, that doesn’t mean that other people who are less grounded in the realities of making accounting decisions don’t think it’s possible.  This article in Forbes about a company called ScaleFactor, that was able to con investors out of $100 million dollars for its promised AI accounting system, is very interesting.  I have to admit that I had never heard of this company before seeing this Forbes article.  Surprisingly, their website is still active; so the Forbes expose hasn’t shut them down yet.

Reading the Forbes article, describing how the company used actual human accountants in the Philippines to perform the shoddy accounting tasks while claiming that their magical AI software was actually doing it, reminded me of the extremely interesting documentary I saw not too long ago about Theranos, called The Inventor: Out for Blood in Silicon Valley.  That company claimed to have invented a one drop blood testing machine that could quickly test for a huge number of things, including cancer.  It turned out that the machine didn’t work, but they pretended it did and had human lab workers run tests, often erroneously, in their “fake it ‘til you make it” scheme to con over $700 million from gullible investors.  Paying powerful celebrities for their endorsements also helped convince people that it was worth sinking their money into.  They literally bought credibility because they couldn’t earn it properly.

As always, we should try to learn from these kinds of mistakes that other people make so that we don’t need to suffer the same kinds of losses that they had.  The “fake it ‘til you make it” approach to start-up companies isn’t new by any means.  Sometimes they actually work.  I remember attending meetings in Silicon Valley where some successful entrepreneurs actually bragged about using that tactic.  I have no statistics on how many of the fakers actually made it to their goals and produced real products and services.  I would hazard a guess that for every one that did make it, there are dozens of fakers who didn’t make it, often crashing in flames.  Investors, which include employees of companies promising “magical” things, should be extra cautious.

Sorry, the comment form is closed at this time.

 
%d bloggers like this: