This month I had the privilege of presenting at the Great Lakes Software Excellence Conference. The theme of the conference this year was “AI.” We focused on ways industries are embracing artificial intelligence to help companies make better decisions. The conference was attended by people from software development, logistic companies, hospital networks, food distributors, among others.
My topic was Predicting Corporate Behavior Using Data Analytics. I presented on the use of data analytics to recognize patterns in corporate behavior. A graphic artist, Laura Leenhouts created a large graphic poster of each presentation. As they say, a picture is worth a 1,000 words. I was amazed at how well she captured my remarks.
Mary Brown, an inclusion consultant, reminded us of how our bias affects the design of machine learning software. For example, she described the bias in human resources software used to screen job applicants. The training data used to teach the AI may have a gender or ethnicity bias based on the training data used. Tamara Faber-doty, VP of IT Consumers Energy described her company’s theft detection AI application that recognized over 1,400 incidences of abnormal electric usage that triggered further investigations.
There were some funny examples of AI mistakes in Janelle Shane’s Lessons Learned from AI Gone Wrong. For example, when shown several training images of fish, the AI software recognized the fingers holding the fish rather than the image of the fish. Is it a mop or a Sheepdog? AI has a hard time telling the two apart. Her talk made me realize how difficult image recognition (including facial recognition) can be.
The Google representative, Gary Coburn, demonstrated several Google tools, to help us create our own machine learning application. Google has created an AI Chatbot that can make restaurant reservations on your behalf.. I was impressed by what Google offers for those of us inclined to experiment in the world of AI.
Lastly, were we cautioned by several speakers, including Ryan Johnson, Lead Engineer at TwistThink, to be sure to start our AI projects with a clear, compelling need in which the solution can by quickly validated to show a return to the budget owner. Questions he proposed to consider included: Are we solving the right problem? How is our bias limiting the outcome? Is there reliable data from which to train the AI?
All-in-all the GLSEC 2019 was a good example of how regional conferences can help local companies up their game with AI tools and applications to stay competitive in our rapidly evolving business world.