Covid-19 is accelerating the shift away from human call center operators and towards AI-powered chatbot assistants. As customer service bots are becoming more common, they’re also becoming more realistically human, displaying the sort of empathy and consideration necessary for a good customer service experience. But AI’s ability to simulate empathy could put real people, and huge numbers of them, out of work, Bloomberg reports, as covid-19 speeds up the shift towards job automation in many sectors.

This is particularly bad news for countries whose economies are heavily dependent on call-center outsourcing: Nearly 286k call center workers in the Philippines are at risk of being replaced by chatbots by 2030, according to forecasts by the Asian Development Bank. This is almost a quarter of those employed in the outsourcing sector in the country, which accounts for a whopping 9% of GDP, according to estimates by Oxford Business Group. Other popular outsourcing countries like India, Brazil, and Mexico are perhaps less at risk because of a more diversified workforce.

Egypt itself had made outsourcing call centers a priority: Since at least 2018 and 2019, the CIT Ministry had made growing the outsourcing and call center industry a key policy platform. Back in 2019, the ministry was targeting USD 4.7 bn in outsourcing exports in 2020, compared to USD 3.2 bn. As recently as last December (watch, runtime: 7:34), CIT Minister Amr Talaat has continued to tout the importance of this sector to his ministry, though he did not give any updated figures.

Covid had hit our local industry when it first emerged: While there doesn’t seem to be any official figures out there for the sector’s performance in 2020, the local press noted last April (during the height of covid) that demand for outsourcing and call centers had fallen 70%, with some businesses recouping some of those losses from healthcare call centers.

But others think the alarm is overblown: Chatbots are still a long way away from being able to replace humans for more complicated queries, some experts say. Developing the software systems that run AI chatbots in the first place, as well as maintaining them and monitoring their performance, will also require human labor, thereby creating more jobs in a slightly adjacent sector.

Some think the goal should not be to replace human empathy but to augment it: Machines can be taught “cognitive empathy,” says author Minter Dial, who has written on the subject, by learning to assess the context of the situation and how that might make a human feel. But they cannot be taught emotional empathy, which is the simple co-feeling one experiences if another human is happy or sad. For that reason, a human being may still need to be part of the customer service process, perhaps after first having the customer screened by a chatbot. In looking to create empathic AI systems, we should not confuse the appearance of empathy with its actual existence, says AI researcher Stuart Russell, adding “we may even want to reserve the realm of interpersonal relationships for humans.”