
A Silicon Valley start-up has created an AI-based solution for one of the lingering questions faced by the customer service industry: How do you make sure that your call centre employee, who is likely to be based in India or the Philippines or another non-English speaking country, is clearly understood by your customers in the US or UK? Sanas’s real-time voice-altering technology converts a range of accents into a more “neutral” sounding one for smooth interactions between employees and customers. But is that really a problem that needs solving?
In a hierarchy-free world, a software that tones down or erases a person’s accent to make it easier to understand might have been a great idea. As such a world is still a long way off, however, the software only ends up raising troubling questions about cultural and linguistic identities. For example, who decides what is a “neutral” accent? Just like two Indians from different parts of the country don’t have the same accent, an American from, say, Boston and one from New Orleans would not sound the same. The creators of the software argue that it removes some of the burden currently placed on call centre employees who have to undergo months-long training and makes them less susceptible to the ire of customers. But the voice samples shared by the company online sound flat and robotic — perhaps an inadvertent admission that achieving the kind of “neutrality” envisioned by those who would erase differences rather than acknowledge and celebrate them, is not even humanly possible.
In a sense, the software is an extension of the syndrome represented by Apu Nahasapeemapetilon, the improbably-named Indian immigrant convenience store owner on the American animated series The Simpsons, whose manufactured accent (performed by a white actor) has long offended many in India. Is wiping out difference that much better than the mockery that was Apu’s “Welcome to Kwik-e-Mart. Thank you, please come again”?