5 AI & machine learning CRM take outs from SXSW

Share this content
I am writing this as I travel back from Austin, Texas, following my first experience of SXSW. For those who haven’t been, SXSW is an eleven-day conference and festival celebrating the film, music and interactive industries and aptly describes itself as the ‘destination for discovery’.
 
As head of data and digital operations at Armadillo, I'm naturally interested in what others in the industry are thinking about artificial intelligence (AI) and machine learning, and the Interactive stream of content at this year’s conference gave me plenty to consider. I designed my festival schedule around panel discussions and seminars where I could listen to the most interesting learnings from a range of global experts. 
 
Whilst I will take some time to fully digest what I learnt and put thoughts into action at the agency and with our clients, here are the five key points which are crucial for consideration by anyone working in customer relationship management that I took away from my time at SXSW, and the events that most inspired me.

1. We're getting distracted by "AI will take over the world" arguments

These arguments don't truly reflect reality. AI is simply a tool no matter how much we anthropomorphise our devices. Using AI devices is only ever as safe as is directed by the humans using them, and the data sets those humans use. Amazon's Alexa as a device can't care what you say in front of it - it doesn't have emotions - but the humans who make the decisions on how Alexa is designed and monetised certainly care. It's up to us to challenge the decision makers to ensure they consider privacy and security and keep them at the heart of everything they do.
 
From “How to build a brighter AI future” by Cassie Kozyrkov, Chief Decision Scientist at Google.

2. Machine learning isn't the magic answer to all of the problems in your business

Machine learning is a very focused solution for a very specific problem. Truly understand the problem you are trying to solve first - is it truly a predictive data problem, or something more fundamental to your business, such as your strategy? Despite the marketing noise around machine learning from all-in-one service providers with expensive annual fees, traditional analytical solutions and a thorough, honest understanding of your business's challenges will often go most of the way to finding an answer for much less effort and time than machine learning.
 
Panel conversation “Data Science Unicorns and Silver-Bullet AI”. Main point made by David Robinson, Chief Data Scientist at DataCamp, along with Randi Ludwig (Dell), David Mesa (NASA), and Caitlain Hudon (OnlineMedEd).

3. To implement machine learning, businesses will need to change how they capture, store and label data

In the future, simply storing data in any fashion in a database won't be enough. Machine learning needs high quality, unbiased data laid out appropriately for it to be effective and efficient. Businesses should plan every stage of capturing and storing data, even if there are no immediate plans to use machine learning for several years. The best data sets for machine learning are those where the captured data is designed for that purpose in the first place. 

4. We need to think about what we want from our AI assistants

Think about the last time you were stressed. Most of the time, you probably didn't mind if someone in the office or at home reacted to your stress and hopefully helped you feel better. But how would you feel if your phone, car or even your toaster noticed and responded to your mood, even if it was for your benefit? We're not quite there yet, but it's something we will be confronted with.
 
Panel conversation: “Inside Story of Building AI & Tech for Real Humans”, Main point made by Anna Pickard, Head of Brand Communications at Slack, along with Wally Brill (Google), Ed Doran (Microsoft), and Andrew Hill (Daimler).

5. Don't forget that humans - your customers – and not data are really at the centre of machine learning

All of us are going to be affected by machine learning within the next few years, if we aren't already (just think about your smartphone keyboard...). The decisions and actions your business make using your customer's data will have real-world implications, and will reflect any problems inherent in your business' data, whether it's accidental bias due to way data is captured or the decisions you make that influence the predictive outputs from machine learning. Take care to use machine learning responsibly and not simply as an excuse for your actions, and continue to ensure your company is transparent with your customers about how you plan to use their data, now and in the future.
 
Though we were thousands of miles away from Bristol, the topics that arose from these discussions all lie close to home, and all are concerned with focusing on the key priority in any AI or machine learning discussions: the customer. 

About Nicholas Blake

Replies

Please login or register to join the discussion.

There are currently no replies, be the first to post a reply.