After hefty cuts, AI puts thousands more telco jobs on the line
If you are training to be a line judge in tennis – which presumably involves long hours hunched over, hands pincered around knees, staring at a strip of chalk as your eyes smart – you might want to reconsider. A sports footnote this past week in the AI-claims-jobs saga is that men's professional tennis plans to drop them like a dud ball starting in 2025. Hawk-Eye, the computer-vision technology previously used for disputed calls, will replace human judges entirely, as it did briefly during lockdowns. It's game, set and match to AI.
This move will percolate sideways and up in tennis officialdom, if not down, where line judges remain a rare, expensive luxury. The women's game will follow, as it invariably does. Umpires, who may have risen through the ranks of line judging, will fear an AI changeover atop the laddered chair (they are also rarely used outside the biggest events). Meanwhile, ball boys and girls will eye Spot the robot dog, used at a Madrid tournament this week for security, with trepidation (because robot-dog security guards aren't scary enough). An update with mandibles could put them out of summer jobs.
Few will cry for the tennis officials made redundant. Their 15 minutes of fame usually comes when they make a mistake. Half of them look like they would struggle to pick out the top letter during an optician's eye test, let alone determine if a small, fuzzy ball traveling at up to 140 mph has nicked one millimeter of a faded line. They won't feature in many lists of jobs lost to AI.
Those lists are lengthening, though. This week, IBM's CEO Arvind Krishna reportedly announced a hiring freeze while speculating that 7,800 jobs (a worryingly precise figure) could be replaced by AI in the next few years. This is only 3% of the total at the bloated diplodocus of the tech sector (no consolation for them). But at least Krishna isn't parroting the usual piffle about automation "freeing up" employees for other tasks or AI creating new roles.
It is finally dawning on many people, if not all the world's democratically elected governments, that Hawk-Eye, ChatGPT, GitHub Copilot and other permutations of AI threaten a jobs apocalypse. Goldman Sachs issued a report in March predicting AI would "replace" 300 million jobs and citing the recent impact of generative AI.
The telecom sector looks extraordinarily exposed. For one thing, it's stocked with people in sales, marketing and customer services, including high-street stores increasingly denuded of workers, like those coffee chains where you select your beverage on a giant touchscreen instead of telling somebody what you want. Chatbots have already replaced some roles. One very big operator is known to be exploring the use of ChatGPT in customer services for added efficiency – a move that could turn thinned ranks anorexic.
Outside research and development and construction, the network is also, basically, just a very expansive and sophisticated machine. The whole aim of technology officers these days is to minimize human involvement, the rationale being that humans break things and make the sort of fat-fingered mistakes a slick AI would not. Expressions like "zero touch," "closed loop service assurance" and "self-driving network" were circulating long before GPT was even a twinkle in Sam Altman's eye.
In other words, when a telco is not digging up roads or abseiling down poles, its network could feasibly be a self-operating, self-healing entity, stripped clean of people, run by an AI that's probably been developed by Google or Microsoft even though it lives in facilities owned by the telco to keep GDPR watchdogs and other regulatory authorities on side. All those fault-monitoring, trouble-ticketing and other routine technical jobs have gone. If staff have been "freed up," it's not to do other jobs at the telco.
How this might look in the interim is something like Rakuten. Better known for its ecommerce activities, the Japanese company claims to operate a nationwide mobile network with an operations team of just 200 people. An equivalent "legacy" operator would normally have between 6,000 and 7,000 in this area, said Rabih Dabboussi, the chief revenue officer of Rakuten Symphony (a subsidiary selling telco products), at a recent media event. Automation is treated like a religion. When Rakuten started building a network a few years ago, its managers were issued strict commandments. Thou shalt not hire more people topped the list.
Thou shalt forever cut jobs, seems to be the edict many older companies have pursued. While several of the big telcos tracked by Light Reading have yet to publish figures for the recent fiscal year (ending in either December or March), the exodus of staff continued at most others, whose collective headcount fell nearly 58,000 last year. Across AT&T, T-Mobile and Verizon, the big three of the US mobile market, around 45,000 jobs disappeared in 2022, more than 11% of the end-2021 total.
Outside the US, around 11,000 jobs went at Deutsche Telekom, Orange, Telecom Italia and Telefónica. That was a much smaller 2.5% of the earlier total, and yet more than 67,000 non-US jobs have been cut from the payrolls of these companies since 2018, a figure equal to 13.5% of headcount at the end of the previous year.
Much of this attrition has very little if anything to do with technology. Instead, it's the result of more routine efficiency measures and the disposal of assets, including geographical units, infrastructure (such as towers) once but no longer deemed strategically important, and IT resources farmed out to the public cloud. This is a frightening thought for employees. If jobs were disappearing this fast before the arrival of ChatGPT, what does the future hold?
Cheering on Rutger
Further layoffs, would seem to be the obvious conclusion. Weeks ago, Microsoft's telco team was highlighting a tweaked version of GitHub Copilot – a sort of ChatGPT for writing software – that would be able to code for the unique requirements of a telecom operation. Ericsson is suddenly pitching network applications (rApps, in the lingo) that can power basestations up or down at 15-minute intervals depending on local circumstances. People would never have done that anyway. But apart from writing apps and digging (and the software is starting to write itself), what's left for living organisms?
The usual retort is that AI remains fundamentally stupid, whatever the advocates say. It can produce a Harvard-level essay after barreling through the libraries of the Internet but has no understanding of causality or any "general intelligence." This is a good response to the argument that AI will eventually rise up, kill us and then live in our homes, watching Bladerunner and cheering on Rutger Hauer. But AI doesn't have to resemble a gun-toting Arnold Schwarzenegger to threaten livelihoods.
The best hope for many people is that companies will keep them around as a kind of civic duty, like communal benches in city centers. Deceased anthropologist David Graeber reckoned many jobs already fell into this "bullshit" category years before the ChatGPT revolution. Orange's policy of not replacing every employee who retires is clearly an answer to the problem of having a unionized and/or civil-servant workforce, impossible to shrink fast. But it also suggests these people have not exactly been integral to the smooth running of Orange.
Many occupations could pass through the tennis line-judge phase, where employees show up and continue to feel needed while a more capable bit of software hovers patiently in the background, intervening when human frailties are seen. Thousands may be trudging off the court in the next few years.
- Microsoft dangles generative AI for telcos and slams 'DIY' clouds
- Be afraid – ChatGPT has caught the eye of telecom
- Telcos don't look very brutally automated
- ChatGPT prompts excitement and fear about the AI future
- What the rise of the robots means for BT
— Iain Morris, International Editor, Light Reading