The Dawn of Agentive Technology: Chris Noessel on the UX of “Soft” AI
Header photo credit: Jess Myra on Twitter
As part of Interaction16, in a packed room at Finlandia Hall in Helsinki, Chris Noessel gave a fascinating and compelling talk on the dawn of agentive technology, and the implications for UX designers.
The audience provided a mixed response to the opening question, “Who is afraid of Artificial Intelligence (AI)?” Chris elaborated from his experience with students that perhaps the question might be, “Who is afraid of what humans will do with AI?”
Chris talked about the emergence of a new category of technology that works on behalf of users to complete tasks. This category, which he is calling agentive technology, can be seen as a particular form of Artificial Intelligence.
Artificial Intelligence can be broken down into three categories:
- Artificial General Intelligence: For example, HAL from 2001 A Space Odyssey. This is the AI we are most used to thinking about, and see in sci-fi – a computer intelligence which would be like that of a human brain.
- Artificial Super Intelligence: For example, Samantha in the movie ‘Her.’ This intelligence would far surpass that of humans. It is hard for us to comprehend what this intelligence would be like. The ‘singularity’ refers to the moment that this intelligence comes into being, and the world is forever changed.
- Artificial Narrow Intelligence: For example, Whopper in War Games. This intelligence can learn and infer but cannot generalise. Narrow intelligence exists today.
Chris then breaks Artificial Narrow Intelligence down further:
- a) Assistive Narrow Intelligence: This technology helps you in real time to complete a task, for example tax software like Turbotax.
- b) Agentive Narrow Intelligence: Agentive technology does things on your behalf, while allowing you to turn your attention elsewhere. This is an emerging category of technology, which will need new approaches to user experience design.
Examples of Agentive Tech
The Roomba vacuuming robot automatically vacuums your floors, navigating the space, returns to its charging dock when needed and cleans according to a schedule you set.
Get Narrative is a wearable camera that automatically captures images by default every 30 seconds. A user can clip it on and go, and entrust the job of capturing moments to the device. The images are stored in the cloud, and the smartphone app allows users to browse their images. Image algorithms do the work of highlighting the best shots.
Giving the user a promotion
In both of the above examples, technology works on behalf of the user to get something done – in these cases, cleaning the house or capturing memories, while the user gets on with their life.
Chris talked about the concept of giving the user a promotion – users get promoted to managers of task do-ers. The potential power of agentive technology is to free up time by offloading tasks. Agentive technology can act like a butler or a valet – working in the background to ensure everything’s running smoothly.
Of course, this comes with some questions about what tasks we should be offloading to technology. Chris identified the potential for narrow AI to help us with jobs, connecting, physiology, skills and art. He also talked about the tasks that we have to do, but might be unwilling to do, and the tasks that we have to do but we cannot do alone.
What does this mean for UX?
From a user experience standpoint, agentive technology brings new design challenges. When dealing with an agent, there are several key interactions to consider:
- Setting up the agent – how does the user learn about how to manage the agent, setting goals, doing trials, all with low risk and the ability to step in when needed?
- The user needs to learn how to stop the agent.
- Monitoring the agent – how does the agent alert the user to emergencies or recognize the trends that pre-empt a negative outcome?
- How can users play with the agent’s capabilities, tune the behaviours and handoff when they have established significant trust?
Chris is in the process of writing a Rosenfeld book on the topic, which is due out this year. In the meantime, if you want to hear more, this episode of the UX podcast features an interview with Chris.
The ongoing development of AI, and in particular Agentive Technologies brings new possibilities for UX design. Agentive tech delivers on the user centred promise of helping users to complete their tasks quickly and efficiently. With this, comes a new set of principles and understanding that designers need to develop. It’s a brave new world, and for it to work well, UX needs to be at the forefront.
Towards an agentive future…
Agentive tech relies on a different cycle of engagement, where users may disengage from a technology once they have set it up to their satisfaction. This comes with risks, as we hand off agency to the agent.
For those who would like to learn more on this topic, Chris is working on a book about agentive tech with Rosenfeld media, which is due out in 2016. In the meantime, this episode of the UX podcast features an interview with Chris on the topic.
Chris convincingly articulated the need to think carefully about the implications of AI and the new skills that designers will need to have to create these systems. Agentive technology delivers on the user-centred dream of providing the user ways to complete their tasks with reduced effort, but it is not without risk. Humanity needs to have a careful, thoughtful hand in the creation of all types of artificial intelligence.