Voice tech expert Joan Palmiter Bajorek discusses bias in AI, misconceptions about the voice tech industry and why she started a nonprofit organisation for Women in Voice.
As a self-described “recovering academic”, Joan Palmiter Bajorek joined the tech industry because she wanted to make an impact.
Her most recent role was head of user research at AI start-up NLX, where she worked on UX design, data science and products. In 2020, she was ranked as an industry leader by Voicebot.ai and she is also the founder and CEO of Women in Voice, an organisation that aims to celebrate, amplify and empower women and diverse people in the voice technology field.
Bajorek has worked on conversational AI products across various industries including healthcare, telecoms, utilities, transport and customer service. She describes voice tech as being “at the intersection of impact, tech and language”.
“I’ve always loved languages,” she told Siliconrepublic.com. “As I pursued research in linguistics, I found the field of phonetics and speech recognition software, which demonstrated how acoustics and speech data were already starting to revolutionise interfaces as seen in early Duolingo and Siri products.
“I stumbled upon the field in 2015 and knew this was the direction of my career since then.”
‘When we incorporate voice tech into immigration, job hiring and autonomous vehicles, we need to be extremely cautious about what our datasets look like’
– JOAN PALMITER BAJOREK
Bajorek said that when it comes to working in the voice tech industry, it’s important not to set user expectations too high, as it may lead to users dismissing voice tech as a whole when it is a field that is still developing.
“The real magic is when we can create delightful, personalised experiences that support users fully.” She said that while working at NLX, she was helping to build “user-centric and data-driven products” that leveraged conversational AI in a multimodal way using voice, visuals, audio, video and touch.
Voice tech works off AI and augmented information that can leverage patterns found within datasets. However, while AI offers many benefits and innovations, it can also lead to inherent bias being built into systems.
Bias in voice tech is a field of study Bajorek began looking at in her graduate studies and one of her PhD papers was published in Harvard Business Review in 2019.
“When we incorporate voice tech into immigration, job hiring and autonomous vehicles, we need to be extremely cautious about what our datasets look like and bias we may accidentally train AI on,” she explained.
“We need diverse teams building inclusive products and this is one of my main drivers to create Women in Voice, the international non-profit I founded in 2018.”
Bajorek said Women in Voice now has more than 100 ambassadors across 20 chapters in 15 countries. “Women in Voice amplifies, celebrates, and supports diversity in voice tech so that we can continue making more diverse teams across the industry,” she said.
Outside of the concerns around bias, the voice tech industry also faces concerns from users around security and privacy. Bajorek said that voice tech data requires security and privacy measures just like with other types of personal information. However, she thinks there needs to be greater transparency and understanding of the technology given to everyday users to avoid any misconceptions.
“While concerns about privacy are valid, the biggest splashes I’ve seen in the media about this were primarily design errors rather than ‘Big Brother’ type of behaviour,” she said.
“One of the most common questions I get about voice tech are related to privacy and how intelligent Siri and Alexa products are or aren’t. However, I’d love it if we could shift the conversation to what voice can enable and the boundless opportunities in this space.
“Voice tech can increase accessibility to everyday products, create transformations in customer service, save critical time in healthcare settings, and much, much more. Imagination is the limiting factor for much of the practical work we can build today.”
Updated, 5.23pm, 10 March 2021: This article was amended as Joan Palmiter Bajorek recently left NLX. A previous version of the article stated that she still worked at the company.
Want stories like this and more direct to your inbox? Sign up for Tech Trends, Silicon Republic’s weekly digest of need-to-know tech news.