Professor Verena Rieser is one of the co-founders of Alana and our resident expert in anything to do with ethics and combating bias in Conversational AI. Verena recently received a Leverhulme Trust Senior Research Fellowship from the Royal Society in recognition of her work in developing multimodal Conversational AI systems. Here we ask her to share a little bit of insight into her ambitions for the future of AI at Alana.
How would you describe your role at Alana?
My official title is Head of NLP. Natural Language Processing, or NLP as it’s often shortened to, is a subfield of artificial intelligence that deals with how computers can understand and produce language.
A lot of the work we do at Alana happens in research laboratories. But not everything we design in our labs is ‘fit’ to be used out in the real world by customers. A big part of my role is working with the team to make sure what comes out of the Alana lab has a realistic (and responsible) application in the real world. It’s very important to me that Alana helps to build a positive role for Conversational AI in society.
Is there really a role for ethics in Conversational AI?
I believe so, and we’re certainly building ethical behaviour into the capabilities of Alana. As Conversational AI technology advances, our interactions with voice assistants and robots are becoming more natural and human-like. This has highlighted a number of ethical implications connected to the widespread use of Conversational AI technology. This is my particular area of interest and it’s a responsibility we take seriously at Alana.
For example, a lot of the latest state-of-the-art techniques in NLP are based on machine learning. By learning patterns from very large datasets, these machine algorithms can also inadvertently pick up undesirable societal biases. For example, we know that some big AI models are negatively biased towards gender roles, thinking that all doctors are male and all nurses are female. My ambition is to make sure our systems are responsibly built, especially since some of our applications are targeted towards vulnerable user groups, including children and the elderly.
What inspired you to develop a career in Conversational AI?
I’ve always been fascinated by language and how people hold conversations. My first area of study was literature and linguistics. I was also an active member of my university’s theatre group and I interned at a radio station.
Almost by accident, I stumbled into a statistics class during my undergraduate degree, which I hugely enjoyed. This led me to apply to programmes teaching computational linguistics. I went on to earn my master’s degree and my PhD in this subject area.
What do you think will challenge the status quo of Conversational AI in the near future?
I’m personally very excited about three things: Multimodal Conversational AI, ethical questions around AI and NLP, and models that combine symbolic and deep learning approaches. I can see these are all going to be important disruptive forces in Conversational AI in the coming years.
Does Conversational AI play a role in your family life?
My children are two and five years old and they are both already ‘voice interface natives.’ It’s amazing to see how easy it is for them to operate the TV by using their voices or to select their favourite music. That being said, I often have to switch off the microphone, otherwise we would never have a quiet moment.
Was there a fictional AI that captured your imagination before you entered this field?
I grew up in the 1980s – an era when everyone watched Knight Rider and dreamed of driving a clever car called KITT. I saved all my pocket money so I could buy a KITT one day.