Breaking the ice for designing with AI
Baby steps for AI-infused dashboards in Cognos Analytics
Background
About 3 months ago, I got the opportunity to be part of an AI design facilitator group, lead by Jennifer Sukis (Design director of AI transformation at IBM) to train to specialize in Entreprise design thinking for AI.
As part of my self-assigned missions of my education, I started to look for opportunities for AI activation in my team through facilitating education and AI design with my team members.
I am part of the Cognos Analytics design team. Cognos Analytics is a business intelligence platform that is used for discovering, analyzing, visualizing and sharing data. It is a complex and ‘old’ product, which means that there are quite a big number of smaller teams that are working on the different components of it.
This includes a specific team that is responsible for infusing AI services into the product. Even though I know there are reasons behind the structure of the teams, this specialization of an AI team has resulted in many Cognos designers not feeling comfortable working with AI.
This presented a great opportunity for me to try to break the ice for designers outside of the AI to become more empowered to use AI in their design process and to develop my AI design facilitation skills.
The first team I approached was the authoring team. The authoring team works on the experience of creating data visualization dashboards and reports. It’s one of the bigger teams who have a lot of user research and a good understanding of their user journey, so they seemed like a good candidate to start experimenting with my skills and the workshops with.
My plan was to start small, to gain more confidence and test out the EDT activities and how they flow and build-up to longer, more specific workshops with bigger and more diverse participants
Data-driven dashboards workshop
The first workshop we did was around data-driven design. Not all the team is co-located so we used Mural for the workshop.
Goals
The goals of this first workshop were simple:
- To identify opportunities for gathering data from dashboard users that would help the team deliver on their hills or KPIs
- To start the conversation with the authoring team around infusing AI in different parts of Cognos
- Help me start to become more comfortable with explaining AI concepts and facilitation AI design workshops*
The plan
In the first 5 mins, I introduced the training group that I’m part of and explained goals nr. 3 to the group. It was important for me that the team knows that this is all a prototype so that we can work together on improving the workshops and making the activities valuable to them in a way that fits their process and needs.
Next, I spend some time talking about data-driven design: What the benefits and risks for it are and how data-driven design can be valuable to our users and our process. In addition, we went through some explanations of the different types of data that can be collected and what it can be used for.
The 3rd step is what too the bulk of the time of the workshop. This is where we went to through the actual activity of identifying the data that can be collected from users to design a better and faster authoring experience.
There are different templates for running the activity. We used the one shown below. However, in retrospect, it might have been better to start with a simpler categorization for introducing thinking about collecting data.
It was interesting to observe how the majority of ideas were in the user data column, this is likely because a lot of the ideas for faster dashboard authoring can be around personalization.
We played back the ideas and discussed a few of them, but mostly we discussed if the activity helped in understanding what it means to collect data for data-driven design.
Afterwards, we talked briefly about the tools we can use at IBM to collect this data (Platform analytics). And In the end, the team was asked to give some feedback about the workshop as well as what they think would empower them to work with AI more and better.
What worked well
As the first workshop, I was quite nervous running it and was sure there will be a lot of things that will not go as planned. However, I was happy that in terms of addressing the goals it had a few successes:
- AI activation: Starting the conversation about AI design in the team and breaking the ice about it being a complex thing that not everyone can understand is obviously the biggest victory for me.
- Data is not scary: We started the data activity with lots of unsure looks and comments but decided to dive in and try it anyway and it started to seem simpler the more ideas were added and the team expressed positive feedback about understanding even a tiny bit more about data-driven design following the workshop
- Learning about team needs: the feedback grid, as well as the discussion, was super helpful in understanding what the team is struggling with. My biggest takeaway was that the team was interested to learn more about what they can do with AI and learning to identify opportunities for AI design. Which lead to the next workshop.
Learnings
- Imposter syndrome: As someone who is still relatively early in their journey in learning about AI and practicing design facilitation, this was a big struggle. However, I can only imagine that more and more practice is the way to get over it.
- Confidence in explaining concepts: My biggest realization was that understanding something and explaining it to others are totally different things. More reading, more practice and more conversation
- Workshop prep: Examples really help in understanding AI concepts, they were missing from this workshop but I learned I need to make sure to have them in future ones.
Identifying AI moments in the authoring journey
A few weeks later, based on the feedback from the first workshop I ran another workshop with the same team, although we invited more members from authoring now and took 1 1/2 hours this time.
Goals
The goals of this workshop were to address the team’s need to:
- Get a better understanding of what AI is good for, and how it can be used
- Identify opportunities in the user journey where AI can add value
The plan
With a similar structure as the first workshop, I used the first 30 minutes to explain what AI can do and go deeper with examples of AI capabilities.
For this part, I used the many resources and links prepared by Jennifer Sukis.
Then, and again for the majority of the workshop time, we worked with the user journey that the team had already prepared through their user research to identify AI opportunities.
The journey phases, steps and pain point had already been pre-populated before the workshop. The whole team was very familiar with it already since they have been working with it for a while.
We were only interested to see if we can use a user journey and pain-points to identify AI goals and focused on that lane in the user journey activity.
After we spent 15 minutes individually brainstorming ideas, each person got a chance to vote on 3 ideas that they would like to explore further.
Due to time constraints, we only proceeded with one idea to explore and agreed to explore the others with the presence of more stakeholders who could validate the ideas in the next workshop.
Takeaways:
- Involve a bigger team: I have to admit I still did not feel confident enough in explaining AI concepts to stakeholders who might be experts in the field, and before the workshop, we were all hesitant to include more stakeholders and wanted to test out the activity first.
However, by the end of the second workshop, both the team and I felt that we would have liked other stakeholders to be part of the discussion. I consider it a success that designers would ask for that since it means a) they think the workshops are helping them expand their ideas for improving the experience b) they (like me) are starting to feel more empowered to work with AI with developers and data scientists who might have more experience and knowledge about AI - Identifying opportunities: the activity definitely helped the team think of AI solutions to address pain and friction points in the user journey. Some of the ideas that emerged can be considered “small usability enhancements”, while others need to be addressed as a broader design epic outside of a specific step in the user journey.
I’m interested to explore ways to make an AI design workshop flow to support both directions. - Questions: the biggest questions that came up after the activity were around the next steps after identifying opportunities and how we can push to add AI design solutions to the product roadmap and make it happen.
In addition, as well as trying to understand when the right time for an AI workshop would be.
Next steps
Not only is there still a lot more work and opportunity with the authoring team, but there are also many other Cognos team that I’d like to start running workshops with.
