An Inside Look: Making AI Unbiased at Solstice FWD
Sarah Berger, Marketing Manager
Artificial intelligence (AI) is changing everything, but at the same time it changes nothing.
We see mentions of AI in every tech article, newsletter, press release, tweet, cover story — it’s top of mind in our industry as well as our business. Still, the rise of AI in social networks like Facebook, Twitter, and Slack is slowly raising the expectations of customers in terms of prioritizing the information we care about.
What hasn’t changed? The customer-centric methodology to yield worthwhile outcomes from emerging tech. It is still true that AI projects that begin in the innovation lab — without clear business value targeted at the outset — are doomed to fail. It is also still true that early validation with customers is the best way to determine if further technical validation is worthwhile. After all, if customers do not see the value in a company generating insights for them — whether using machine learning or not — the juice is not worth the squeeze.
On June 21, we’re bringing a human-centered design approach to AI to life at our annual digital innovation summit, Solstice FWD. We wanted to solve a real problem, and do it iteratively, by following our CXDD (customer experience driven development) methodology to allow our customers to validate the experience and prioritize the features they prefer.
The result? An AI experience born from Solstice Labs, called Brainwave, that uses computer vision, sentiment analysis, and machine learning to help attendees network with others at the event.
But rather than taking my word for it, I recently sat down with Solstice’s Principal Digital Strategist and product owner of Brainwave, Jared Johnson, to discuss how AI will impact every business — and the world at large — as well as what you can expect to see from our team at FWD 2018.
Q: What is the most common misconception about AI?
A: First, AI is a suitcase word — people cram all sorts of meaning into the AI acronym. AI is often thought of as approximating human intelligence, but it’s not close to that yet. Second, the same fears that people have today about AI have existed for the last 50 years. Third, just because you can use AI doesn’t mean you should use AI. AI can have a lot of value within your business, but only if it is targeting specific outcomes. However, if you can accomplish the same outcome using explicit programming or linear regression, AI can be overkill for your needs.
FWD sneak peek: Author of Our Final Invention, James Barrat, will demystify AI and explore its complexities, future potential and its dark side.
Q: There is some debate about the positive versus negative aspects of AI. Elon Musk has even warned of an “AI apocalypse”. How can we co-exist with intelligent and super intelligent machines in a way that’s ethical and for the common good?
A: We are such a long way off from needing to worry about AI apocalypse scenarios — they are thrilling to talk about, but simply not a worthwhile concern in 2018. A real ethical concern for AI today is about how we are using data to train AI or machine learning systems without human bias being baked in. For example, computer vision for facial recognition originally did not work as well for black people because the first training data was mostly white people’s faces. If we are codifying our own biases (conscious or unconscious) into our training data, then our AI will be just as biased. The more we can be conscious of our own biases, the less biased our AI-based systems will be.
FWD sneak peek: Google’s Reena Jana and Divya Tyam will deliver a dual-talk at FWD on eliminating bias and designing for inclusion to make a better products for everyone.
Q: How disadvantaged are companies who are not using AI?
A: AI offers topline and bottomline improvements to your business. If you are not using AI, your competitors will find ways to reduce their own costs so they can be more profitable, or offer lower prices. Enterprises can also use AI to differentiate themselves by offering valuable experiences to their customers, like insights or predictions that customers will pay more for.
Q: You’re helping create an AI experience that recommends other FWD attendees to network with by using computer vision and text sentiment analysis. How did you come up with this concept and why do you feel like it’s important?
A: We listened to our attendees from 2015 and 2016 and learned that executives want to network and meet like-minded people who are facing similar challenges or driving similar initiatives in their organization. The idea for this experience was born from wanting to solve a problem: Networking at conferences can be overwhelming, and finding the few people in a sea of attendees that you will jive with in a mutually beneficial way is really difficult. We wanted to reduce all that noise to a simple suggestion of the top three people we think you should meet.
Q: Explain how you’re getting data — like tastes, mood of writing on social media, and career habits — from attendees?
A: We are using a connected camera to look at your face while you watch some videos to see how you react. Do you smile at a Nirvana music video? Did you frown during the clip from “The Godfather”? All of this information helps us understand your tastes a little better. The tone of voice you use on social media is a good indicator of your personality type as well. This information helps us assess you from a personal and professional standpoint.
Q: How can the results of this AI experience help businesses?
A: All of the AI components and services that we have combined into one place each have value to businesses. You do not need to construct a system as complex as ours to gain value. You simply need to identify the outcome you are driving towards or the problem you want to solve and test different hypotheses on how you can achieve that outcome.
Q: With the recent Facebook data breach, some may be wary of sharing personal info to the masses. How are you ensuring this data you extrapolate is secure?
A: We are following best practices to securely store the data. We’re also just storing the data required to recommend FWD attendees to one another. When we recommend you to other attendees, the only data being shared is your name, company, and LinkedIn photo (if you logged in to your FWD portal with your LinkedIn). None of the data collected will be used outside of this experience.
FWD sneak peek: Our leading keynote, Laurie Segall from CNN, interviewed an emotional Mark Zuckerberg on the recent data breach.
Want to learn more about how enterprises can leverage AI? Email Jared Johnson at email@example.com. Interested in attending FWD? Register at solsticefwd.com or shoot me an email at firstname.lastname@example.org.