(Montreal) Although we’ve been hearing more about its potential pitfalls and risks recently, AI can also contribute to securing individuals’ data.
Some financial institutions use speaker recognition, that is, they authenticate a customer using telephone services by the distinct characteristics of their voice.
Desjardins implemented the technology in June 2021. Since then, 1.5 million members, who previously consented to the process, have been identified by their votes in six million calls.
Authentication process [traditionnel] Our members called him a nuisance when they called him, says Annie-Claude Jutras, Senior Director, Transformation of Customer Relationship Centers at Desjardins. It took several tens of seconds at the start of each call to complete all steps of the process. With voice authentication, it becomes faster, more natural, and more fun. »
This method is also safer, says the master.I Jutras, where Desjardins uses a type of passive authentication, in which the computer system recognizes the voice during a simple conversation, and that of a hundred parameters. “We model the sound with these parameters which give us a unique key for each organ, in the form of an alphanumeric code,” she explains. It’s like a fingerprint. »
By doing this, the customer does not need to mention a passphrase to be identified, a type of authentication that has already been used by scammers, who unknowingly record a person’s voice to impersonate them and gain access to personal information.
“This kind of authentication happens in an automated system, Nuance MI Jutra. Training an artificial voice or using an AI engine to calibrate the voice to perfection is much more challenging in the context of a fluid discussion where questions and answers are exchanged between the member and the Desjardins employee. »
Laurent Charlin, Mila Senior Academic Member, Associate Professor at HEC Montréal and Canada-CIFAR Chair Holder in Artificial Intelligence, agrees.
“We humans can remarkably recognize the interlocutor’s voice, but the system can be trained to distinguish very specific tones,” he says.
“If some financial institution deployed this system, they must have made sure that it was better than the technology they were using before,” he adds.
Fake it is not as easy as you think
According to Patrick Cardinale, professor and director of the software engineering department at the Higher School of Technology (ETS), technology does not yet make it possible to simulate an impromptu conversation with the voices of others from a simple speech sample.
“The fact that authentication takes place during a conversation creates uncertainty about upcoming questions or exchanges, so it is difficult to predict the answers. Moreover, to get something reliable, it would take a huge amount of data to train. [l’intelligence artificielle] “Most people don’t have that data online,” he comments, referring to the audio clips that are available online.
It’s not the same for TV, radio, and film professionals, whose audio snippets are all over the web.
“We have already seen videos of actors or politicians saying things they never said, thanks to hyperfakes (deep fakes), Laurent-Charlin confirms for his part. There are also applications that, from small snippets, can generate your voice by making you say what you want ».
In case of doubt, it is always possible to refer to traditional methods, such as identification questions, recalls MI Jutra.
Checked with the Canadian Anti-Fraud Centre, no fraud has been reported involving artificial intelligence with voice reproduction, in particular for gaining access to victim bank accounts. “However, this is something CAFC will continue to monitor closely,” Geoff Horncastle, Acting Communications and Client Engagement Officer, said via email.
This post was produced with financial assistance from the Meta Exchange and The Canadian Press for News.
“Music guru. Incurable web practitioner. Thinker. Lifelong zombie junkie. Tv buff. Typical organizer. Evil beer scholar.”
