I wouldn't trust them to post a letter
AI. BUSINESS. The Post Office, Volkswagen, the Tay Bridge and AI: matters of faith; not in technology, but in people
“I wouldn’t trust them to post a letter” is a common enough statement in English. It is an expression of someone’s inability to rise to even the most basic levels worthy of another’s faith: to actually do a simple act that is asked of them. The Italian Post Office has, in its time, been unworthy of the confidence placed in it to post, not letters but postcards: sellers thereof would always offer an envelope along with their picture postcards, because the Italian Post Office had the somewhat offhand habit of simply burning them, once, of course, it had first secured the money tendered to it for the stamp. One ought ordinarily to be able to trust a post office to post a letter. That’s not the question here. What was in question in Britain’s case was its ability to charge and account the correct amount for the stamp.
In 1999, the British Post office introduced a new accounting system for its branches and sub-post offices, called Horizon. A sub-post office is essentially a private business, which may offer services and products other than those purely of a postal nature, and they tend to be family businesses, involving a husband and wife, or other partnerships, due to the heavy workload they generally entail. The new accounting system was to have made life easier for them.
In hundreds of cases (approximately 736, in fact), the new software did not make things easier but actually made things a lot more difficult. Shortfalls in takings started to be revealed, and the Post Office demanded that these shortfalls be made good, by the people who were running the businesses. Sub-postmasters and sub-postmistresses found themselves having to pay out to the Post Office sometimes many thousands of pounds to correct their inefficiencies in handling cash, a skill that many had learned by the hard graft of counting individual pound notes and pennies over a good many years.
Some of them got into such difficulty that they were sent to prison; others withdrew from the business: they simply could not cope with the large sums they were required, on the orders of courts and their senior Post Office management, to reimburse to the Post Office. Their names were dragged through the mire and some of these pregnant thieves (as one was dubbed), who are generally held in a position of very high regard in British society, as custodians of a vital communications service, lost their homes, their livelihood, their standing and were even publicly vilified in the press and on the streets. One at least took the only way out that was left to them: they committed suicide, out of shame at their inability to match the precision of the software that told them that they were so inadequate.
So numerous were the cases of financial impropriety, and so numerous became the court cases raised by the Post Office to exact their pounds of flesh, that public interest in the matter spread from just being at a local level to a national level. An inquiry was started into the issue in 2020, and the Post Office staunchly defends its ground by saying that its software was irreproachable. And yet, it was far from irreproachable.
The courts have found that there were serious faults within the software; that much of the money that had been demanded of sub-postmasters and -mistresses over the years had simply not been due — albeit in sums of 30, 40 or 50,000 pounds; at the very least, there was no cogent reason to believe the software was any more accurate than the post office’s own sub-postmasters had been. Some who had stood up for their rights received monetary compensation; some have been released from jail; some refuse to contribute to the inquiry, on the grounds that it is a whitewash. Indeed, officials at the Post Office received bonuses and promotions, and went on to take up lucrative positions at other organisations. It is truly one of the most outrageous scandals and miscarriages of justice to have ever involved a British public institution. And it arose fundamentally on the basis, either, of a belief that a computer can never get it wrong, or, of a strategy that a computer is so commonly held to never get it wrong that, even when one knows that it is wrong, it bolsters one enough to brazenly make the assertion that the computer is right.
On a windy night on the last Sunday of 1879, the Tay Bridge, owned partly by the then North British Railway and built as a huge distance and time-saving structure to span the estuary, or firth, of the River Tay, south of Dundee, in Scotland, collapsed into the sea. The whole centre section sheered off in a violent gale, taking the train and its up to 75 passengers and three crew with it. None survived.
The bridge was a major civil engineering structure and, by contrast with the two other stories related in this article, no computer played any part in its design, erection or use. Thomas Bouch was a renowned engineer and had had a number of successes and had earned himself thereby a favourable reputation. However, his reputation led him, as we say, to rest on his laurels, and he failed to take into sufficient consideration the bedrock of the river, down below water level, on which the bridge should stand, or to effect quality assessments of the mass production work that was mostly carried out to produce the iron pillars that supported the trackbed above water level; it was mass-production work that had precisely been factored in to make his offer as chief of works the most attractive, and therefore the one to win the letting of the contract.
It was as a result of inadequate foresight, an underestimation of the forces that would put pressure on the structure, and failings in the quality of the structure itself that ultimately led to the disaster. When, ten years later, works were in full course for the construction of another bridge across another estuary to the south of the River Tay, the Firth of Forth, an entirely different design of bridge was deployed. It is a magnificent cantilever structure that still stands to this day and is renowned for having a full time team of painters, who start at one end at the beginning of each year and work their way across the bridge painting it until they get to the other end at the end of the year, then returning to repeat the exercise, ad infinitum. It is strong, safe, sturdy, and it has stood the test of time for 130 years now.
If you look at photographs or films of the Forth Bridge as a train is crossing, what is remarkable is that the train is barely discernible. So massive is the structure of the Forth Bridge that the mere possibility of collapse is put beyond all imaginability. The Forth Bridge was not designed to have this sturdy structure in order just to carry trains from Edinburgh to Burntisland; it was so designed in order to actually induce people even to board the trains that plied the route. Public faith in railway safety had been dashed to such a degree in the light of the Tay Bridge disaster that only gargantuan structures like that over the Forth could restore confidence. The Forth Bridge is a wonder of engineering, no question; and it is a wonder among public relations statements.
In the 1990s, in Belgium, Germany, Britain and many other countries across the world, a revelation was in the making. Concerned at the pollution of city air by motor vehicles, regulations were being introduced to test, and tax, vehicles that produced a large quantity of pollutants.
Europe is a continent whose executive employees had, by that time, become used to being granted the luxury of a company car, and, in light of the revelations from tests done on vehicles, it was found that the least polluting vehicles were those that were powered by diesel engines. This revelation came as a downright surprise to those who had observed dumper trucks and steam rollers chugging up the countryside’s lanes puffing thick, black, acrid smoke out of their exhausts.
Our governments, however, were reactive to this strange news, which, besides vehicle pollutants, was emitted from testing stations across the continent, and indeed other continents. Duty on diesel was reduced to make it an attractive fuel type. And, in a widespread change of policy, diesel became the norm for the company car. Whole fleets switched to the fuel in reaction to the fantastic news that diesel was so benign to our health and that of the planet. It was seen as the ecological thing to do to go diesel.
Many years after this switch in mentality had taken place, news started to emerge that the pollutants emitted by diesel vehicles were indeed very much on the low side. However, they were on that low side only for such time as the vehicle was actually being tested. It was discovered that some vehicle manufacturers, notably Volvo and Volkswagen, had built software into their increasingly computer-driven automotive components, which quite simply issued false readings whenever a vehicle was undergoing emissions testing.
Governments, citizens and competing vehicle manufacturers were outraged. These car makers had colluded to trick entire nations out of their peace of mind, their health and their environment, had effected a wild distortion of the car market and all this in a bid to gain an unwarranted upper hand in the automotive sector. Very soon, diesel vehicles became a taboo subject: they were banned from city centres and those who, in good faith, had complied with their government’s policy to go diesel, now had, at their own cost, to abandon diesel in favour of the next big thing.
In an especially pernicious case, one Fleming who regularly drove his wife to hospital check-ups in Brussels, only a few hundred metres within the newly introduced “low emissions zone” found he was no longer able to drive his spouse to the hospital, and this was a great inconvenience. He sold his old car and purchased a new one so that he could comply with this new health-giving regulation, which was nevertheless having detrimental effects on his wife’s health. In their new car, the couple were delighted to be able to drive, as they had before, to the very door of the hospital.
Shortly thereafter, the city’s administration decided to relent, given the difficulties posed to persons of age who owned older vehicles and could not afford to change them. They kindly re-designated the few streets around that hospital back outside the low emissions zone. No doubt those who, meantime, had bused or walked or biked to the hospital were pleased at this act of understanding. But the couple who had changed their car were less comprehending: they had incurred no end of expense, using funds they did not have, to accord with the rules as government had decreed them. They did not appeal the rule, they took it on board and suffered in order to ensure compliance, as a good citizen does. They were not so delighted: they were speechless. At the utter lack of foresight by the Brussels government, at their insouciance at creating a social problem, all in the name of solving a social problem, and then simply changing their mind after the fact, as if that were the encapsulation of administrative magnanimity.
These three cases, in which unshakable faith was placed in an object of revered trust, two of them being computers and one a man of unshakable engineering reputation, sent shockwaves throughout the relevant societies. Government policy was even structured around the findings made by computers, and engineering standards were blown out of all proportion, all in order to establish public belief. Those who placed their faith in Horizon and Volkswagen had been wrong to do so.
The British Post Office and Volkswagen diesel scandals, and the Tay Bridge disaster, are all salient, headline-grabbing cases; the Brussels low emissions zone has grabbed fewer headlines, but I myself unwittingly crossed the line one evening and ended up with the bill: for 15 minutes driving around Watermael-Boisfort at three in the morning, I was served up a demand for 350 euros. Rules are rules, of course. And that computer didn’t get it wrong, though I swear I did not see the “stop” sign for my car.
Everyone is chatting about ChatGPT, and it is a wondrous thing that these men and women of wisdom have created: a machine that can write like Geoffrey Chaucer. Perhaps it could even have saved a great deal of argument and nail-biting by simply polishing off Gustav Mahler’s Tenth Symphony, which musicologists have anguished over since the dear man passed a hundred and more years ago. Perhaps it could even have discerned the errors committed by musicologists in attributing multifarious works to Johann Sebastian Bach, which, it has been established, were in fact written by other composers. Perhaps, even, an essay by a student at the university, who contends that it is his or her own work, could be fed into AI to see, indeed, whether it has previously been published by another. For, if AI can grind out essays within minutes, it surely knows the essays it has ground out and can therefore attest to the requester’s attempted fraud?
But what if I did write something entirely of my own imagination and then that got fed into AI? What, then, if it flagged my work as having already been published by another? How solid a defence could I put up against that? As my standing is hacked down and public scorn is poured upon me. It took an enquiry to discover the defects in Bouch’s iron. And it took 20 years to discover the defects in Horizon, and about the same to discover the mendacity of Volkswagen. How right would the computer be deemed to be in my case?
The British Post Office has not taken any significant measures against the officers who pled blind, even smothering plain evidence to the contrary, that their software was beyond reproach. Sir Thomas Bouch died in ignominy ten months after his bridge collapsed. He freely admitted to his failings, for which 78 lives were ended. Apologies have been forthcoming from motor manufacturers Volkswagen and Volvo, and others involved in the diesel fraud; but still, and, it has to be said, to a large extent, rightly so, there remains widespread faith in the computer, just as there remains widespread faith in the originalty of written work and civil engineering structures worldwide; and there is similarly largely justified faith in the computer’s ability, when directed to process appropriate material, to always come up with the right answers. GPO, Bouch and VW: which of them was most culpable in abusing that kind of faith? And which, in the end, was least dishonourable, when they were found out?
With the advent now of AI in such a widely accessible form, another question arises. Scepticism was levelled at the abilities of computers, back at the time of their genesis, to do anything approaching what they can now do. It’s accepted today that they have grown far beyond what even Alan Turing, their inventor, could have imagined. But, might this scepticism yet turn tail and ultimately cause us, in our lassitude and indolence, to question the abilities of humankind to even produce original, reasoned, correct work, and, instead, to adopt a kind of faith in computers that becomes but blind, whether through ignorance, inability or machination: so that it comes to be cited as evidence by people seeking advancement on the backs and efforts of creators of originality? Will we ultimately have to defend our originality against a standard that is set by a plagiarist in whom we are exhorted to place our faith, our entire faith: one enshrined in a technology that, just like man himself, is all too malleable to the whim of him who seeks gain?
To that extent, one can conclude, indeed, that, when we bow to AI as a faculty superior to man himself, then, at that point, if not before, man and machine are on an exact par with one another: each will bend as willingly as the other to the whim of a controller. And that is without considering the fact that we ask computers to do what is almost physically beyond the abilities of man: such that he has to trust the computer, for the cases show how long it takes him to prove the computer was wrong the whole time. Enter Franz Kafka.
In the meantime, whilst there is little doubt that artificial intelligence has, in some circles, been roundly lauded and knighted as the quintessence of our new age, I feel safe in warranting that this new force, even without a coronet on its countenance, is unlikely, like Sir Thomas Bouch, to have, ten months from now, quietly passed away, hidden from public view, in the rural idyll of Moffat.