Software developer Ibrahim Diallo had a nightmarish experience last July. When he tried to walk into his office, nestled within a Los Angeles skyscraper, his security pass did not allow him in. A friendly guard did, but it was not the end of the story. Many doors on the inside denied him entry, and so did his computer. Critical files disappeared as his computer login failed; the system rebooted automatically and, finally, revoked his access.
What started as a slightly bizarre tech error, led to apocalyptic reactions as nothing could be done to get Diallo another pass or restore his access. Not even his bosses could overrule the system although they assured everyone that he was still employed. The incident escalated to a three-week work outage because an outgoing manager had failed to mark Diallo's contract as an ongoing one in the new system. (At the time, the company had been acquired by another and was in the throes of a transition while Diallo was just eight months into his three-year contract.) The entire chain of processes, triggered by one error in data, led to a situation that could not be reversed. As per computer record, Diallo had been fired, and a couple of days later, he was frogmarched out of the office by security personnel. By the time his system-generated firing could be sorted out, he had given up and moved on to another job and the company lost an expert who was passionate about his work. Diallo wrote about the entire incident in a blog post, a timely warning to all companies which depend too much on 'dumb automation'.
According to experts, a comprehensive artificial intelligence (AI) system using mixed techniques could have prevented such an erroneous outcome. To begin with, a smart system is built on encapsulated knowledge (domain expertise coded up) and thrives on deep learning algorithms - artificial neural networks which emulate the human brain, learn from large datasets and recognise patterns. Based on Bayesian updating and fuzzy logic, AI can reanalyse situations and revise its reasoning as new information comes to light. Also, there are more checks in place and a more balanced approach to ensure 'deep understanding' of nuanced incidents. Armed with these features, a 'True AI' would not have done an actual firing without flagging several reasons. On the other hand, a system without adequate technology features is likely to goof up when it has to deal with human errors and get to the bottom of it.
Nearly a year after this unsettling incident, e-commerce giant Amazon is now under fire for assessing and firing thousands of warehouse workers in the US. And once again, computer software is doing it. If productivity and quality of output fall below the threshold, workers are terminated. In spite of media reports, Amazon denies that firing takes place in this manner and says that managers are entirely free to intervene. This may be true, but managers may not do so, leaving people at the mercy of the software even though they missed targets for entirely understandable reasons. Amazon has drawn flack over the years for treating its colossal staff like robots rather than human beings. In fact, companies all over the world are increasingly using AI-powered software programmes to assess workers, implement workplace surveillance, and hire and fire people, undoubtedly a dangerous trend as we rapidly move towards automation at work.
Rude and Reckless
Dockless and app-enabled electric scooters are all the rage nowadays, but several riders in Brisbane were in for a cybershock when they rented Lime's connected scooters. These bikes have voice boxes that connect to the Lime app, can be picked up by a renter and left at a drop point when the ride is over. However, people were taken aback when some of these e-scooters suddenly started shouting sexual comments and other offensive things. In fact, an entire row of stationed scooters could be heard yelling lewd and racist comments, creating an unsavoury cacophony. Undoubtedly, it was some hackers' idea of an amusing prank - they hacked into the scooters and changed the audio files. A number of videos featuring the shouting scooters have been posted online by media houses and members of the public. Lime vehicles are currently on trial in Brisbane as the city council is considering whether to grant it a licence to operate permanently.
Lime is currently checking its entire fleet to see how many had been tampered by "vandals". Before the voice box hacks, the transportation firm had to update its software to fix a glitch that made its e-scooters brake suddenly when riders were going down at full speed. Although the hacked vehicles will be fixed soon, the recent attack is an ugly wake-up call concerning how easy it is to hack connected devices and what kind of havoc they could wreck.