Sneak peek into the “Design Thinking for AI” course & curriculum
In fall 2020, I was invited to teach a small course on Design Thinking for AI as part of the Business Innovation program that I myself graduated from. The 3rd-year BA students were well acquainted with the design thinking methodology and had little to no knowledge of AI (beyond the headlines) or other design disciplines such as UX or interaction design.
Thinking up the Design Thinking for AI course
I had done plenty of talks around the topic in the aftermath of launching the AI meets Design toolkit in 2019, but this was the first time putting it all together into one program. Having a total of 5 classes of 3 hours together, I considered which building blocks I thought fundamental.
My objectives were for the students to walk away from the course:
- Having a basic understanding of what Artificial Intelligence (AI)/Machine Learning (ML) is
- Knowing current common capabilities and how to spot and validate their potential for innovation
- Becoming aware of UX considerations for AI/ML-driven products
- Reflecting on the potential positive and negative impact of widespread AI/ML on society
I used this as an outline and designed a session around each of the objectives above.
Guest lectures & Individual class projects
I decided early to invite guest speakers for short lectures to make sure the students had deeper insights and more diverse perspectives to learn from and luckily they all accepted!
I wanted to make space for personalization to make it more relevant and apply the learning. After the introduction, the students were asked to select an industry/topic/SDG/challenge that they were personally interested in and would develop a use case around throughout the course. With these 3 elements and a whole lot of slides, we got started.
The program was 90% remote so we used Teams andGoogle Slides.
Sharing my program
In this article, I share an overview of the curriculum and approach I created for the course. I hope it serves as a resource for anyone in a similar position on either end. For students (enrolled in an institution or autodidactic); to work through some of these materials, and exercises, and educate themselves, and for educators; to exchange practices, inspire, and build on each other’s work.
Class #1: Introduction & Crash course
Setting the stage
I believe setting the stage is key. Before diving into the material, we mapped what the students already know about AI, their motivations, and what they’re hoping to learn.
Fundamental technical concepts
Together with Kwan, we covered fundamental concepts to equip the group with a basic understanding of what Artificial Intelligence(AI)/Machine Learning (ML) is — and isn’t. This included terminology like a machine, deep, supervised, unsupervised, and reinforcement learning illustrated with well-known examples such as AlphaGo, SPAM filter (the OG of algorithms), and Spotify recommendation (my personal fav). Kwan, speaking from her experience as a software engineer and working on computer vision systems, walked the students through all the steps in the ML development and training process, from data collection to evaluation. We critically reflected on AI as a tool, asking what are its limitations and where does it truly add value, concluding that indeed:
“Essentially, all models are wrong, but some are useful”
— George E.P.Box
We discussed the complexities and the role and value of non-technical experts in the process. Thank you Kwan for joining the session and sharing your in-depth technical knowledge & contagious passion for tech ❤
Dissecting a familiar system
Building on this new vocabulary of concepts, the students picked apart an AI system that they interact with regularly. How does it learn? From what data? How does it arrive at and present its predictions? What is it optimizing for? How do we evaluate its performance? In small groups and with the support of Google, they shared their insights and speculations.
Homework: Research existing applications in your domain
Students were asked to pick a challenge they wanted to work on and research 3 related AI applications. For each, specify: What does the product/service do? How does it use data and ML? What’s the added value? Then pick one and dissect it using the Machine Learning Canvas by Louis Dorard.
As prep for the next session: watch this video from Google about the ML process.
Class #2: Applied AI
After sharing and reflecting on the homework as a group, this session introduces the realm of applied AI and how to spot and develop a good use case where AI can add social, user, and business value.
For this, we welcomed Benjamin Flader, an innovation consultant focused on tech and AI projects. We shared 4 approaches to opportunity spotting with AI and starting points for each:
- user-centered — what are your users’ biggest pain points?
- data-driven — which data is available within the organization and publically?
- business-driven — what are the biggest costs and how might you alleviate employees from repetitive and under-stimulating tasks?
- tech-driven — which building blocks are available and which new developments are taking place?
We familiarized ourselves with current common capabilities such as computer vision, natural language processing, and pattern recognition so that students could recognize ‘easy wins’ and take advantage of available API’s and off-the-shelf solutions.
“AI shines in problems where the goals are understood, but the means aren’t” — Yonathan Zunger
Ideation being all about the number of ideas, in the next phase we aim to assess the ideas on the design thinking trinity :
- feasibility — is it technically possible? How complex is it?
- viability — does it make business sense? What are the costs and returns?
- desirability — does this help people? Who and how?
And the 4th criteria which I like to add:
- responsibility/sustainability — what might be the wider impact of bringing this idea into the world?
Thank you Benjamin for sharing your expertise and giving the students a little peek into consulting work ❤
Homework: Ideating and selecting AI solutions
As homework, we sent the students off to come up with their own ideas, offering my AI ideation card deck as a brainstorming aid. They came up with 20+ ideas each, plotted them on a value/investment matrix, exchanged peer feedback, and picked one final idea to continue with. They wrote a 100-word paragraph evaluating their idea from a user, technical, and business perspective and filled in a simple template to boil down its essence in 1 page:
As prep for the next session: browse through Google‘s People + AI Guidebook and read 1 chapter.
Class #3: UX of AI
In the 3rd session, we talk about how to design the user experience and the front end of your AI-centered products/platforms. We looked at the areas of Trust & Transparency, User Autonomy & Control, and Value Alignment which I wrote extensively in this article series based on my ebook with AWWWARDS.
Avantika Mohapatra came by to share how user research can support building a solid understanding of user needs and context. By moving the technical solution to the background, we can ground design decisions about the system in the experience of people that are ultimately interacting with and affected by them.
If you aren’t aligned with a human need, you’re just going to build a very powerful system to address a very small — or perhaps nonexistent — problem.
— Josh Lovejoy
Thank you, Avantika for sharing this urgent reminder and offering pointers on how to integrate these practices into the process ❤
Homework: Develop your idea
Students were tasked with the next steps in developing their idea. First on the user side, by talking to people about the problem they’re solving, iterating on a clear & complete value proposition statement, and taking notes on how each of the UX challenges we covered shows up in their product. Then technically, by doing 1 hour of minimal tech research for useful datasets, pre-trained models, tools, and APIs they could potentially use as building blocks
As prep for the next session: watch Josh Lovejoy’s TED talk “Fair is not the default” and thinks of 3–5 pitfalls for bias, unintended consequences, or negative impact of your product/service.
Class #4: Impact on Society
Time has come to zoom out and reflect on the potential positive and negative impact of widespread AI/ML on society, and the importance of considering bias, inclusivity, transparency & ethics, and thinking through the unintended consequences of our innovations.
Bias, inclusivity, ethics & other fails
We looked at various types of bias and how to anticipate, recognize, and reconcile them. While offering suggestions on how to manage these within projects, we acknowledged how the root of these problems is not technical, but founded in systemic inequalities and toxic narratives. Likewise, we underlined the risk of overreliance on ‘good intentions’ and the need for accountability. We rounded this up by looking at a handful of classic AI fails that received major backlash, discussing what may have happened and how it might’ve been approached differently.
Applied AI ethics
In the final 30 minutes, we invited Celine Dutier to share her research on and framework for actionable AI ethics to bring home the urgency and complexity of this perspective in AI projects. Thank you Celine for showing us around your expansive and deeply-researched thoughts and Miro board around the topic ❤
Homework: Consequence wheel
Using the consequence wheel as a template, we invited students to speculate on the potential unintended consequences of their solutions through the STEEP (Social, Technological, Ecological, Economical, Political) lenses. They developed strategies and made changes to their system and process to mitigate these risks and navigate them mindfully.
As prep for the next session: Reflect on your learnings. What is your greatest fear when it comes to AI? Biggest hope or potential? How do you see your role in this?
Class #5: Wrap & reflections
The 5th and final class was almost entirely dedicated to reflection and integrating the learning. We discussed key take-aways and looked back at the mind mapping we did in the 1st session to see how far we’ve gotten in our understanding of AI since. I am really proud of this group and impressed with their critical questions, quality of their work, and most importantly: their curiosity, willingness to learn, and overall good vibes. Thank you all! ❤
We wrapped with feedback and testimonials of which some snippets below:
If I were to do the course again (in a similar context), I’d spent slightly more time covering the fundamentals and include more moments for peer review, feedback, and non-structured Q&A with myself and the guest lecturers.
It’s a wrap! I hope you enjoyed reading about this program as we did doing it. If you’re a student or educator (or human of other sorts), I’d love to hear your thoughts and what you got from this piece :) Thank you for reading.
If you’re looking for more of this, sign up for updates on my online course I’m looking to launch later this year, come check out what we’re doing over at AI x Design, or reach out for similar or custom AI x Design programs.
AIxDesign is a place to unite practitioners and evolve practices at the intersection of AI/ML and design. We are currently organizing monthly virtual events, sharing content, exploring collaborative projects, and developing fruitful partnerships.
To stay in the loop, follow us on Instagram, Linkedin, or subscribe to our monthly newsletter to capture it all in your inbox. You can also find us at aixdesign.co.