By Chris Gilmour and John Fraser
People are currently terrified of Artificial Intelligence (AI) – and so they should be, as we have seen and suffered from the impact of its forebear, Artificial Stupidity (AS).
The dangers of AS were brilliantly highlighted in the TV series Little Britain, now sadly regarded as very un-woke, which featured a series of sketches with comedian David Walliams in drag sitting in front of a computer screen.
In each sketch he would be approached with a simple request, the sort of thing a well-behaved computer could accommodate in nanoseconds.
And each time, he would type something, check the screen, and respond: “Computer says ‘no’”.
It was clear from these spot-on sketches that AS was in command, held all the cards, and was intent on being as difficult as possible.
On arrival at the offices of some of the country’s biggest firms, you may already be required to sign in via a computer screen. In our experience, it is vital to have a human on hand as well. You guessed it – because the computer loves to say ‘no’.
The theory is wonderful. As these more and more intelligent devices extend their capabilities, they should be able to speed things up, save on labour costs, boost efficiency, and enhance our lives.
In reality, it just doesn’t work like that.
On a recent visit to a branch of PostNet, the aim was to courier some papers to a government office in Europe.
Step one involved filling in a form. Names, addresses, ID numbers, and contents of the envelope. Had there been spare space on the form for inside leg measurements, no doubt those details would have been required as well.
The form having been completed, one should have been able to pay and go. Not a chance. Not without the computer getting in on the act.
A PostNet person took the form and then proceeded to enter into the computer each and every detail that had already been written down, and not very speedily. This gave the opportunity for some further complications.
What is the phone number of the recipient? No idea!
The computer said ‘no’. It needed a phone number, or the package would not be accepted. So, the computer (with a bit of human help) had to Google away to find the number. This took more time.
Had there been a faster and more logical way of gathering the info, and loading it into the system, frustration levels would have been lower and far less time would have been wasted.
Makes sense? The computer said ‘no’.
Possibly the most annoying examples of AS come whenever you need to contact a call centre.
Instead of a human being speedily answering your call, finding out what the call is about and then putting you through to the right person, instead it is answered by a computer.
You are guided (often unintelligibly) through a series of stages, each involving a series of choices. Enter 1 for…. Enter 2 for… And so on.
Then the concert begins. As call volumes are high (are they ever not?) the caller is forced to wait and wait while often unenjoyable, poor-resolution, music is played.
Possibly it is intended to calm you down. It doesn’t.
Now and then there may be a recorded message, reminding you of those high call volumes, blaming Covid, the weather, God or Satan for your frustration.
Last year a call from Pretoria to the Virgin Atlantic call centre took one and a half hours of waiting, only for the connection to be severed by the airline.
Another attempt was made, another hour and a half dragged by, and the call was disconnected once again.
It would be unfair to suggest that this customer disservice is unique to one virginal venture. It happens almost everywhere, almost all the time.
Try conversing with the robot-run WhatsApp service of Discovery Health. A nervous breakdown may result.
Or how about the newspaper subscription you took out and now want to cancel? There are British publications that make the quest for cancellation more impossible to complete than their impenetrable crosswords.
Of course, machines can work wonders. If you are eventually put in touch with a health care provider, the scanners, X-rays and other hospital machines they can book for you may work brilliantly and can save your life.
AI will advance all this even further. Already, cancers can be speedily and much more reliably detected when computers are brought in to help.
When the tedious and time-consuming side of AI – the coding – has been achieved and all that is left to do is to sit back and enjoy the end result, the effect can be earth-shattering.
Once you’ve entered your biometrics and financial data into the system via the app, shopping at an Amazon or Tesco cashier-less supermarket in the UK is effortless (or so we are informed).
Customers literally walk into these stores, choose what they want from the shelves and walk out. They receive an itemised bill via email shortly thereafter.
The kind of AI technology used in these stores has leapfrogged the radio frequency ID (RFID) chip that was designed to gradually replace bar codes and scanners.
RFIDs are still around but the relatively high cost to manufacture them has meant that they can’t be used cost-effectively in commoditised applications – so they have found their way into luggage tags and clothing applications, where the cost can be more easily absorbed.
RFIDs are incredibly useful for high accuracy stock control of high value-added goods. In a single scan, an operator can measure the entire stock each time, every time.
The learning abilities of AI are astounding, and of course it has many, many benign applications.
However, it is already a concern that whereas some human beings can apply the brakes, make moral judgements and press the panic button when the danger zone is looming, machines may develop their own hostile agendas, and become a real threat.
When that happens, the brilliant humour of the Little Britain computer saying ‘no’ will be replaced by something a lot more menacing.
Combine AI with AS and we are all in big, big trouble.
Chris Gilmour is an investment analyst and John Fraser is a journalist and broadcaster
ZA Confidential 