Human interactions with AI devices can sometimes be a cause of frustration, but what’s really behind this failure, and how can AI services give a better experience?
In only trying to improve their algorithm, many AI-empowered systems are keeping their focus too narrow. For many users, the next step isn’t data being crunched more effectively but a need to be able to guide their device. A recognition that a user's state of mind can change the response they expect to receive. Past choices alone can’t always be relied upon to generate the right answer in every situation.
Everyday AI is a research collaboration between Alice Labs and the Centre for Consumer Society Research, University of Helsinki, in partnership with Reaktor.
The Engaging with EverydAI webinar took place on 5th May at 9 am CET. You can watch back the full research presentation and panel discussion to discover:
Watch the webinar:
Documenting everyday experiences with two algorithmically-empowered services, recommender systems and digital assistants, the Everyday AI research is a vital read for anyone working with AI-systems or looking into their application.Download the research
There are many challenges to getting an AI project off the ground, or refining one already underway.
To assist with this process, Reaktor is making available a series of exclusive downloadable guides over the coming weeks put together by our expert team.
Understanding the impact AI have from a business perspective is a crucial step in ensuring it is sufficiently resourced, in terms of time, money and skills. This checklist of essentials to consider brings together the vital elements and justifications to include when calculating the value and impact of a project and organising its resourcing.
You are also invited to read further insight summaries and opinions into the full research.
Alexa, why don’t you understand what I want rather than what I say? AI systems are on a constant learning curve to algorithmically improve their replies. Users are generally appreciative of how they respond, but what about when AI gets it wrong? Maybe it keeps mixing up two places with the same name. Or recommending you watch more movies because you watched something for work or study reasons. How can this be solved?
Can AI tell when you’re in a silly mood? Or a sad one? What about when you want to experiment with a new content genre but need a form of curated guidance? A user can ask the same question in two different moods and expect different answers. How can AI services respond to this need?
It’s tempting to see the algorithm as the continuing cure to all AI problems, but it overlooks the importance of how users interface with a service. What happens when a user can see a way to progress a query but the AI can’t? Without the tools to steer, is that a failing in AI or design? Is the next step in AI going to be design-led?