Hey Siri - Who are you talking to?

What does an AI assistant’s voice mean to us? They’re always there and seem to know us so well. Statistics show more than 95 per cent of smartphone users say they use their built-in AI voice-powered assistant, and in 2021 research from Ampere Analysis found that over half of UK internet households now own a voice assistant. These handy personas aim to make our lives easier, retrieving information without even the touch of a button and making some day-to-day tasks more efficient than ever.

The more we trust AI assistants to help us, the more information we give them and the more useful they can become. However, do you consider the data you’re disclosing to the companies behind our helpful friends?

Dr Mark Cote, Reader in Data and Society in the Department of Digital Humanities researches the human in relation to technical objects in order to understand the societal dimensions of data, computation and AI. His project Sentient Beings explores what the voice of AI agents look like, entail, require and take from us.

“Humans have always been technological beings- as we have evolved so has the technology we use and are able to create. However, we are now at a point of radical acceleration, driven by datafication. At the same time, voices are becoming an increasing way in which we interact with digital tools - and by extension, the technological corporations who control them. In this more conversational exchange, what elements of our personal data and private selves might we be giving away all too freely? What exactly are corporations gleaning from our voices?

Large language models are no doubt the most significant developments in computation to date - take ChatGPT for example - and as the amount of data they acquire increases and our AI technology improves, AI assistants will be able to give more and more of a personal response/ prompt. We are beginning to have the ability to tailor our AI assistant’s voice.

The desire for human interaction is what drives us in all aspects of life, we need real engagement even with a technical object. AI assistants provide us this with real time responses. This imposes potential dangers of relationship confusion and privacy concerns.

Amazon Alexa assistant

Our interdisciplinary project Sentient Beings has been developed to address these concerns and to explore what a safer AI assistant could look like. An artistic collaboration offering a new perspective with an immersive and evolving soundscape inviting us to question our relationship with AI assistants, how and where we use our voices, and the value we place on them. We have created a ‘whispering gallery’, where movement and murmuring circulate around you in what seems to be an infinite loop. AI assistants intervene when triggered, asking you how they can help.  This installation invites you to reflect on what personal data we agree to give away, in exchange for convenience in our daily lives, and how much vocal data corporations might have on you already.

We need to assess questions imposed here around consent and whether your perspective changes depending on ‘who’ we think is asking us for information, and if we trust them.

In the near future it may be that disclosing personal information is not only preferable, but required by the company. To access the best user experience, for example the modulation of your AI assistant's voice, may require more than disclosing what you might want for dinner that night.“