The rise of artificial intelligence (AI) voices raises complex questions for society and businesses. How should they sound – and do choices around gender, accents, and speech styles truly reflect humanity’s diversity?
A new report provides an important warning. It shows that these choices go beyond those kinds of societal concerns. They can even put some people at greater risk of falling for a scam.
A study published in Oxford’s Journal of Cybersecurity found that people with certain accents or dialects can be at greater risk of “AI-voice-based deception.” Understanding why is crucial.
But there’s also a flipside. While bad actors misuse this phenomenon, businesses can be an important part of the solution – reaching more people, delivering a better customer experience, and removing the opportunity that scammers may try to abuse.
Understanding the risks
Every day, AI voices are becoming more realistic, nuanced, and natural-sounding. It’s no surprise that 21% of voice actors say they have knowingly lost a job to a synthetic voice.
The more human they seem, the more persuasive these voices can be. Fraudsters use them to make calls and try to trick people into thinking they’re speaking with a real person. Anyone can fall for this kind of scam. But some people are at greater risk than others.
Oxford’s study on “vigilance toward AI voices” put people to the test. It had them listen to audio clips of voices and asked them whether they believed the voices were human or AI. Importantly, the study included participants who speak a certain dialect and AI voices that match their dialect.
The research uncovered that when people with a certain dialect heard an AI voice that matched the dialect, they were more likely to believe it was human. Why is that? The answer involves expectations.
Many of these folks haven’t heard technologies speak the way they do. So, when they hear a synthetic voice that does so, they’re more likely to assume it must be real.
The study found that all kinds of listeners “often show a bias towards” assuming AI voices are human. But “this effect is even stronger when the voices use underrepresented regional or non-standard dialects – presumably because listeners are not used to such varieties being represented by speech technology,” the researcher, Professor Neil W. Kirk of Abertay University, explained.
The study explored how to help people avoid falling for this. It found that simply warning people about scams and urging them to be cautious did not make a measurable difference. But something else did: educating people that AI voices can replicate accents and dialects.
When they were specifically given that information, they became far more vigilant. Also, they dropped the default assumption that the voice had to be human.
The lesson for businesses
As more organisations adopt AI receptionists, people will get used to interacting with agentic AI. It’s important to offer them a diversity of voices.
For starters, this goes a long way in advancing the customer experience. Businesses should equip themselves with multiple options for genders, dialects, accents, and a wide range of other vocal characteristics that can be tweaked and adjusted for each customer.
Research shows that people often prefer AI voices that are similar to their own, but not too similar. They prefer voices that communicate quickly, but not too quickly. Each person may draw the line at a different place. So, the best AI call handler for small businesses and enterprises is one that offers flexibility, responding to the way each customer likes to communicate.
The more businesses do this, the more people with different accents and dialects will get used to hearing AI voices that match their own. They’ll no longer think AI tools only speak in certain limited ways. They’ll become less vulnerable to these kinds of scams.
As Kirk explains in his study, certain “language communities” have been “historically marginalised or excluded from speech technology systems.” But it doesn’t have to be this way.
AI tools can be taught to learn, mimic and master all sorts of differences in how people speak. When business leaders embrace that opportunity, they can create happier customers and a safer society. – Inc./Tribune News Service
