AI systems (AIs) are realized by mathematical models and large amount of data, thus there is a lot of research that aims to improve both models and datasets, and, regarding progress in hardware and software computational power, there is a lot of discussion about AIs’ capabilities development. There are two different prospective addressable in Strong AI and Weak AI.
"Strong AI would amount to a “mind” that is genuinely intelligent and self-conscious. Weak AI is what we actually have, namely systems that exhibit intelligent behaviors despite being “mere“ computers."(elements of AI course).
Beyond this theoretical dualism and discussion about machine consciousness, that deserves a dedicated article, as of today we are developing only weak AIs, putting them in various devices. AIs are not autonomous, every output depends on human input, this means that we have to take care of such inputs.
What we should consider as inputs are not only machine learning data inputs but also human interactions. We should care about human–computer relationship, to understand how people perceive AI, how they interact with machine, if they trust AI and how design could improve the user experience.
“It is very important not to remove the human element” says Roger Penrose in AI for good interview.
AI research requires a skillset that have to include user perspective, to meet user’s needs and to create a give and take relationship between system and user. Both qualitative and quantitative design research methods are useful to understand how users are interacting with today's AI systems.
Starting from user research results, the designer role will be to explore and represent user’s needs, but at the same time to understand machine needs designing a communication that enables machine to learn something from the interaction and about the interaction itself.
Such relationship is achievable with a deep human understanding to individuate proper use of data. Data selection, limitation or addiction, have to start from user’s needs, to guide machines in learning directly from user.
Then, why UX and HCI in AI research? AI requires UX and HCI research and design methods to highlight user's problems and to test solutions at any stage.
Contact me at