Can design sprints work for Artificial Intelligence applications?
Last week, for the first time, I attended a meetup on Design Sprints( The Design Sprint Underground)
I had heard of Design sprints from Google – but I am not an expert. The organiser, Eran, created a good atmosphere (as he called it Israeli style interaction ) i.e. oriented to spontaneous discussion – which benefitted the meeting.
I had heard of the idea of Design Sprints from folks at Google and based on the book - Sprint: How To Solve Big Problems and Test New Ideas in Just Five Days
As per Google about Design sprints A design sprint is "a process for answering critical business questions through design, prototyping, and testing" that allows product owners to have "a shortcut to learning without building and launching a minimal viable product". Design sprints have got some traction already
Design sprints involve a five-day process
Day 1. Mapping the problem and pick a focus area
Day 2. Sketching competing solutions on paper;
Day 3. Make decisions to turn your ideas into a testable hypothesis;
Day 4. Create a prototype;
Day 5. Test with real users
Does this process apply to AI applications?
So, my question is: Does this strategy work in building AI applications?
Its tempting to suffix everything with ‘AI’
But that would be superficial.
In a purist sense, if we consider AI as ‘based on Deep Learning’ – and hence, by definition, based on the idea of automatic feature detection from data (typically unstructured data)
AI applications are complex and when viewed in this way, the idea of design sprints have some limitations as applied to AI.
It would be simplistic to apply design sprints to chatbots and claim that design sprints support AI. I believe much work needs to be done. The design sprints idea may need to be adapted to AI (and indeed for more complex engineering applications)
Image source: gv.com