More and more, we’re being flooded by client inquiries about the possibilities of using Artificial Intelligence. While the capabilities and benefits of using AI and machine learning are vast (and growing), this technology should not be positioned as the “magic pill” to solve any and all business problems. But the increased hype serves as a great opportunity to have an open conversation with our clients and help educate them on the advantages of using AI in some specific business cases. Essentially, our clients come to us because they know they want AI, but what they’re really seeking is a simple decision-making engine.
In our previous article “How to Turn Strategy into Success with Digital Transformation”, we briefly touched upon Artificial Intelligence’s ability to help delegate the grunt work and how an organization can capitalize on today’s technology and strategically improve their bottom line. Artificial intelligence (AI) and automation aren’t replacing humans, they’re complementing our soft skills and performing tedious work, like repetitive tasks, analyzing data sets, and handling routine scenarios. Machines amplify our soft skills and collaborate to achieve productivity previously never thought possible. But what is truly possible? Below are The Best, The Worst and The Weirdest examples for the application of Artificial Intelligence and Machine Learning that we’ve witnessed:
- The Best: Programmable. Self-learning. Sensor-driven. Wi-Fi enabled. The epitome of useful AI is in a…thermostat. Nest began by offering its “Learning Thermostat” back in 2011 and then expanding its product offerings to include smoke detectors, security cameras, and other security features. Nest’s popular thermostat conserves energy by optimizing the heating and cooling patterns of both homes and businesses. Within the first few weeks of operation, users manually set temperature controls, and once that data set is established, the Nest thermostat employs a machine learning algorithm to regulate heating and cooling. It learns your family’s or business’ schedule, accounts for local weather patterns, and uses motion sensors to realizes when there’s no one present. All of these variables factor into a predictive and personalized heating and cooling schedule that even shifts to conservation mode, when appropriate, to save energy, time, and money. “Since 2011, the Nest Thermostat has saved billions of kWh of energy in millions of homes worldwide. And independent studies showed that it saved people an average of 10% to 12% on heating bills and 15% on cooling bills. So in under two years, it can pay for itself.”
- The Worst: We all know the saying, “Garbage in, garbage out.” So, any instance where importance is not placed on good data and its best practices can quickly become a disaster. The most glaring of these examples comes when there is the use of bad or incomplete data, i.e. AI should not be used to make judgements on ethics or values. Additionally, that data shouldn’t establish any discrimination based upon race/ethnicity, gender, or even age. One of the most notable examples occurred in 2016 with Tay, Microsoft’s AI chatbot. The Twitter exercise was positioned as a learning experiment where the more users that chatted with Tay, the “smarter” the bot would become. Yet within 24 hours of launching, the endeavor was shut down after taking a disastrous turn – showcased in her flawed (learned and parroted) responses and raising serious concerns about the use of public data to direct AI. As companies are increasingly embracing the move toward machine learning and AI, the risks associated with bad data only become greater.
- The Weirdest (?): You’ve probably been stumped by a ‘Where’s Waldo’ book at least once in your life. But not to worry! Recently, an ad agency, redpepper, based in Nashville, Tennessee built a robot specifically to find Waldo. The bot was created using Google’s AutoML Vision (Auto Machine Learning) tools that were just announced in January of this year. With more than 200K views to its demonstration video on YouTube, the ‘There’s Waldo’ bot is really grabbing people’s attention. The agency not only aimed to showcase the latest tech, but to also test the capabilities of Google’s new service. The process involved a team of researchers cataloging a generous amount of Waldo images (in his many environments) for Google to reference. The robot was then trained to snap a picture of a puzzle book page, and scan every face within the spread to compare against the character. Once found, the bot uses a robotic arm fashioned with a plastic hand to quickly identify Waldo on the page. No easy feat – as some ‘Where’s Waldo’ spreads contain over 300 individual faces.
Whether you’re looking to automate a simple process, like finding Waldo, or build a robust, complex predictive analytics engine based on big data – the first step is an evaluation to determine what technology is appropriate to meet your business need. Within that solution, AI will likely be part of the solution, whether or not the end user realizes it. But you don’t have to adopt AI just for the sake of it. The focus should remain on how machine learning or artificial intelligence can help to solve a business problem, rather than just showcasing new technology. Discover how Rivers Agile can help you revolutionize your operations through the integration of digital technology in all areas your business, radically improve performance and how you deliver products, services and value to your customers – Contact Us today.