By Vanesa Giaconi from Unsplash

Head of UX research tells the head of AI research, “You are a fire in your belly!”

Head of AI research responds, “Can you just be a fly on the wall for just one meeting?”

And I am supposed to be the glimpse and harbinger of the future for these two?

I facilitated the session to create uniformity as a takeaway from an ideation session. Who are our users today, and how might we serve them? We have all this AI capability,we have the most talented designers and AI experts in Washington, DC, everyone is ready and able, the organization (at the highest level) backs the effort — what is left?

Here are the 10 themes (my talking points I used), organized as what I felt would help achieve (the goal) alignment and convergence around research [1].

1. How to better identify and satisfy user needs

— How to conduct interviews, surveys, apply a method for focus groups, and usability testing.

— Once the requirements of the users have been figured out, it is critical to conduct ongoing tests with the users to validate whether the solutions really satisfy their ever-evolving prerequisites.

2. The relevance and significance of user input in AI development

— How to of observing users as they engage with the product — this may assist designers, engineers, and product leaders in identifying what works well [2] and what doesn’t work so well from a usability viewpoint.

— How to examine use statistics to discover how users are engaging with the product. This information may offer insight into areas that require improvement or prospective new features that might be created.

— Does doing A/B testing matter to compare multiple versions of a feature or interface design? Does A/B testing help guarantee that improvements made based on user input truly enhance the overall experience for everyone using the product?

— Continually discussing methods to enhance user experience with stakeholders (such as managers, designers, or engineers) enables a more holistic approach to finding solutions for testing and implementing UX modifications.

3. Efficient methods of data collection from users for the purpose of training AI models

— How relevant are these: web-enabled (digital) surveys; in-person interviews; focus groups; usability testing; and A/B Testing.

— Studies using diaries entail asking participants to maintain a record of their thoughts, emotions, and experiences over the course of a certain amount of time. This information could be examined for trends or patterns.

4. The impact of various interface types [3] on how users interact with AI systems

— Different kinds of interfaces may have a significant effect on the way users engage with AI systems. For instance, a text-based interface may hinder users’ ability to obtain information or accomplish tasks, whereas a graphical user interface (GUI) may assist users better comprehend and operate an AI system.

— The way users understand and engage with AI systems may also be influenced by the interfaces they utilize. When AI has a humanlike appearance (e.g., a face), users may be more likely to anthropomorphize it, which in turn may lead to different expectations about the robot’s capabilities and behavior than if it had a more robotic look.

— The specific design of an interface, including elements such as color choices, layout, and typography, can also affect how users interact with and use an AI-powered system, in some cases making it easier or more difficult for them to achieve their goals, depending on how well these features integrate.

— Users tend to trust systems more when they know more about how they work. Making AI systems more visible (not only physical systems) through their interfaces could help users trust them more in general.

— If we want users to embrace and frequently use an AI-powered system in their daily life, as opposed to giving it a try once and then forgetting about it, we need also think about how simple (or hard) certain types of interfaces are to use and learn. If our users aren’t computer knowledgeable, it’s best to stick to a clear and easy-to-navigate interface rather than a flashy one that only confuses them.

By Nikola Johnny Mirkovic from Unsplash

5. The impact of personalization on users’ interest in an AI system

— Personalized information and experiences may pique a user’s interest in an AI system and lead to deeper interaction.

— An AI system that can learn a user’s preferences could provide faster, more accurate results or suggestions.

— Personalized user experiences have shown to be more enjoyable for users since they are tailored to their specific needs and interests.

— Automated personalization may develop connections between users and an AI system by facilitating better communication and interactions.

6. Do users want an AI system to be transparent [5] and give them control over their data?

— Yes, they want to learn how and why the AI system is making its selections.

— Yes, they want a say in what information is used to train and refine the AI.

— As long as the AI system does its job, users don’t care whether it’s transparent or not.

— Users may prefer an AI system that makes its own judgments without input from them if service recovery is an urgent concern. Users would likely prefer a transparent AI system for more routine activities, such as (example to use) selecting a restaurant to potentially understand why the algorithm made its suggestions.

7. What role does interface design play in facilitating user confidence in and comfort with AI?

— Make the use of AI obvious to users.

— Offer options for customization and control.

— Incorporate user feedback mechanisms into the design.

— Use transparent, easy-to-understand language throughout.

— Optimize consistency across all user interaction points.

8. What ethical concerns [4] must be considered while designing systems?

— Data confidentiality and security.

— User control over their own data.

— Transparent employment and integration of algorithms.

— Potential for misuse or manipulation by third parties.

— Making available technological resources on an equal footing.

9. Under what circumstances do users prefer to assign jobs to AI rather than do them themselves?

Users may feel that AI-completed tasks are more accurate or efficient. Additionally, some users simply enjoy using technology and find it satisfying to let AI handle certain tasks.

— Use case is a detailed report generated from a large data set without spending the time in creating it manually.

— User needs translation services for work but finds hand-written translations too slow and error-prone.

— A user prefers voice recognition software for taking notes because they can speak faster than they can write or type.

— Product recommendations are automatically generated by AI based on their previous search history.

Also:

— When users are confronted with a difficult action that they cannot solve on their own.

— When users wish to automate a tedious process with the help of AI so that they can get it done fast.

— When users want details that are hard to find.

— When AI has the ability to provide a more precise outcome than the user themselves could achieve.

— When AI can do the work more quickly than the user.

10. How do we get more people to use and enjoy AI products?

If we want more users, not just techies, to take advantage of AI technologies, we need to make them more accessible.

— Create advertising plans tailored to AI goods, so that they are more easily discovered and purchased.

–Raise awareness about what AI is and how it may be applied in everyday life via campaigns, seminars, or other forms of outreach.

— Integrate AI solutions in a manner that helps both the platform’s current users and the business itself by working with existing platforms that already have large user bases.

Bonus: with the rise of AI across chatbot use cases, how important is it for an AI system to have a voice or personality that sounds or feels “human”? What kind of responses might users have when they discover that they have been having their conversation with a computer the whole time?