Ethics of IoT and AI
Updated: Mar 24, 2021
I've recently talked in HotMobile 2021 panel on Ethics and Policy of AI Technologies in/for Mobile and IoT Systems. It was a great panel that brought together social and computer scientists for much-needed interdisciplinary conversation.
I only had some images and quotes in my talk (captured in the blog image above). Here is an excerpt of what I spoke about:
When I was asked to participate in this panel, the image that appeared in front of my eyes was very similar to these eggs - fragile tech that needs to be kept in perfect balance. And I will elaborate on this in the remaining few minutes, focusing mostly on IoT at home.
I always believed AI and IoT would go hand in hand because of the original vision for IoT as I understand it. IoT promised a life where "smart" devices are embedded in our environments. These devices collect information about us and "do their thing", and if we base it on the Kevin Ashton quote "without telling us".
I think this was not considered "scary" before because the vision was everyday devices trying to help us, humans, navigate our increasingly complex lives. The focus was on connectivity, and usability, on the technicalities of creating these "helpful" devices. I don't believe there was much thought on the privacy and security of information collection and sharing on the internet.
As technology progressed, we see IoT devices blend more and more into our environments, without standard input/output (like a screen or a keypad), listening quietly, analysing our information and reacting to us or our actions.
The first challenge with this is that IoT at home cannot be considered only for individual use. We share our homes with our families, with vulnerable people, children and the elderly, who do not have the technological expertise to use these devices. Though this statement may be debated looking at my 3-year old :). However, these individuals do not typically have the decision making or purchasing power of these devices. Nevertheless, these devices are part of their living environment, and they learn and transmit data about them as well.
The second challenge is that IoT devices are not only part of our "common" spaces where we are used to sharing with others like living rooms and maybe kitchens. They are increasingly becoming part of our most private spaces as well. With wearable devices, we are opening the doors for even more sensitive data collection, mainly, of our health.
The aim of this talk is not to deny the positives that we receive from our IoT devices and wearables, which do bring several conveniences: we learn more about ourselves, enabling early diagnosis, remote health-care, dealing with isolation during Covid, and so on. But, when I showed these pictures to a group of technologists, designers, and data protection experts in a recent workshop (THRIDI), they told me they found them scary in general.
Still, regardless of how scary things may look to some, we will continue accepting more and more devices into our homes (Joseph Turow quote).
Until this point when we see "Sorry, No Internet Today". The moment people realise that when this happens, nothing works. We all read the news/tweets about people shocked to discover how dependent they have become. Internet outages render their smart devices useless. Their smart devices now cannot even fulfil the functionality of a dumb vacuum cleaner, light or kettle, or they don't know how they can make them work as "normal".
The current state of the operation is too tied to connectivity to the internet, which seems by design. But does it have to be so? Can we give control back to users? Can we make devices that are more transparent about their data collection? Can we allow them to be "smart" with local connectivity, enable users to select when and how to connect to the Internet? Can we allow users to change their usage and data collection practices dynamically depending on the situation, who they are with, where they are and what they are doing? Most importantly, can we do this without overwhelming the users but also with enough granularity for data sharing. Can we find that balance, and here I am going back to the eggs...
This is an optimization problem that needs solving collaboratively with systems engineers, device designers, manufacturers, security/privacy and legal tech specialists, policymakers, human-computer interaction specialists. We do need more interdisciplinary research and development, and we do deserve better than the status quo.
You can listen to the talk in the panel here: