While brands like KEEN have harnessed artificial intelligence (AI) to grow their business, it's still enshrouded in skepticism. We talked to KEEN about breaking down common myths and misconceptions.
Andrew Duffle is a Director of Data Science and User Experience at Portland-based footwear brand, KEEN. He is a champion and early adopter of using machine learning and artificial intelligence in commerce to drive better user experiences. When he’s not leading his team at KEEN, he spends his time on Data in the Raw, a passion project focused on helping companies own their enhanced customer clickstream data to become better, data-driven marketers.
Andrew Duffle is Director of Data Science and User Experience at Keen
At KEEN, Andrew has pioneered new ways of using Einstein for Commerce, and just recently launched a successful holiday shopping page that was completely powered by Einstein Recommendations. Using multiple recommenders, KEEN was able to create unique and personalized gift guides for consumers. In the footwear industry, the scope and variety of shoppers drawn to a brand is vast, and personalization can be difficult. The guides were based on personal browsing history, and brought in the power of dynamic customer groups powered by Data in the Raw. The result was unique combinations of content and recommenders to deliver true personalization and guided product discovery.
While brands like KEEN have harnessed the power of artificial intelligence (AI) and machine learning (ML) to grow their business, AI is still enshrouded in skepticism and disbelief. We asked Andrew for his point of view on the impact of AI, and how he would break down common myths and misconceptions.
Where does AI have the biggest impact in commerce? Why?
Scaling authentic user experiences will be the biggest impact, and honestly, everyone is still trying to figure it out. The concept of using AI in digital commerce is relatively simple if you put it in context of your own personal experiences. Your friends most likely earned “friend” status by listening first, and responding authentically to you, building that relationship and trust over time.
Now, if we equate a friend to AI, or as I prefer, reactive ML, digital commerce will rapidly start feeling more genuine and has the potential to connect emotionally. As everyone in commerce knows, it’s impossible for us to build that relationship one-on-one with users online. This is where AI becomes crucial in the future of digital commerce.
I agree. When AI is done right, especially in consumer shopping, it should feel like a friend recommending products that you trust are exactly your taste. Do you think shoppers will embrace the use of their data to deliver that kind of personalized experience?
Yes, as transparency and effectiveness of AI and ML increase, users will accept it more and more. Right now, we are in a period of industry correction and fear from the handful of data breaches or misuse of data by irresponsible companies. In the coming years, new players in the AI/ML industry will start making user personalization more authentic while increasing the standards of data security. When shoppers start trusting companies with their data in exchange for value, I think they will not only embrace the use of their data, but it becomes the expectation. There is a long way to go as an industry, but it starts with every company owning and being the central hub for their user data.
You mentioned “reactive ML.” That’s a term I haven’t heard before. What does that mean, and how is it different from “AI?”
True AI is rare and rarely done to a level that a user can feel good about being a part of. A subset of AI is ML— and for the purpose of transparency — I am a believer of calling it what it is. Using reactive ML makes this topic very approachable. If we continue with the friend analogy, reactive ML is no different than a friend responding to a movie you recommended. Let’s say you recommend the latest space odyssey movie, and your friend responds with disgust. You most likely will not recommend a movie about a war in space again! This feedback and learning loop make a lot more sense in the context of reactive ML than AI.
What do you see as the biggest myth around AI?
It can solve everything and there is a provider that can do everything. There are tons of problems out there that AI/ML have not solved yet, and there will always be another problem after that.
What are common misconceptions around AI?
That AI and creativity cannot coexist. I believe this comes from fear of AI, and I’d argue the opposite. First of all, AI often automates repetitive and routine tasks, which leaves much more headroom for creativity.
But, more abstractly, solving problems with AI is akin to finding inspiration for a painting. You can use the same kind of thinking to develop an algorithm. One example I love is the Ant Colony Optimization Algorithm. It was written to mimic how ants in a colony optimize the search for food, and AI is applied to work like an ant trying to find the fastest path to a reward. AI can be an exercise in creativity, and it can break down the barriers that prevent you from reaching your full creative potential.
If you had a magic wand, what would you want AI to look like to help you do your job better?
I don’t think I would change AI in general, I would change the negative stigma around it. I would wave a wand and ensure companies start making decisions based on data; this would help ensure AI develops in a way that meets consumer expectations.
Although, it would be nice to have better AI to block robocalls more efficiently. But that’s a different topic for another day.
Learn more about AI in commerce.