Dear Addi P,
I'm hearing a lot about artificial intelligence and that it's being used to get new data insights. What is AI and how does it work?
-Curious in Canton
It's important to note that artificial intelligence isn't really artificial. It's based on real data related to real people, events and facts, and its intelligence is derived from its ability to process massive amounts of information and find patterns, connections and data insights that we might not be able to discover otherwise.
The basic idea is that with a ton of data, really fast processing power, well-designed instructions (algorithms) and effective ways of organizing the data (data models), we can see and learn things that would be too time-consuming or even impossible for humans.
For example, if we want to know what the average commute time for our employees is, we could approach the question in several ways. We could send out a survey and ask employees directly, but it would be hard to know whether their times were affected by such factors as the need to pick up and drop off kids or collect their dry cleaning.
If we're looking at work-life balance, it may be helpful to include everything that happens along the commute so we can understand what that time really involves. If we're looking to compare commute distances and times across the organization, then we probably want to adopt an approach that's based more on an apples-to-apples comparison of distance.
How AI Can Make the Most of Your Data
An AI system could figure out the distance each employee travels based on their home and work location and then estimate travel times using traffic and public transportation data. It would take an extraordinarily long time for someone to work this out without the assistance of AI, but computers can calculate individual and aggregate commute times quickly and efficiently with the right information.
With this commute data in hand, we could compare it with other workplace data to gain a better understanding of how commute times may affect employee engagement, attrition and promotion rates. One might expect people with longer commutes to show higher turnover, but this might not always be the case, as other factors could also be at play. Maybe there's convenient public transportation that makes the trip easier, or perhaps the sales team is on the road so much that proximity to the airport is a more important factor than how far they live from the office. With the right data, AI can help cross-reference these factors to create a clear picture of the situation.
Also in the realm of AI is machine learning, the process by which systems continually work to refine and update their results using new data. This can enable better change detection and increased insight into the connections between data. To return to the example of the commute, if travel times for employees coming from the north suddenly increase, a learning system could determine whether major construction efforts are affecting roads or public transportation.
Understanding when a group of employees is dealing with additional stress can grant the ability to address issues around schedule changes, remote work or even temporary work space more easily.
Making an Impact
One of my favorite examples of artificial intelligence making a meaningful difference is in the area of photo recognition systems. We could teach an AI system to tell the difference between a hedgehog and a badger by giving it lots of photos of each animal (training data), providing instructions and letting it sort the animals' differences by taking stock of measurements, proportions, coloring, patterns, shapes and a range of other factors.
If we then introduce images of dogs, the machine will try to categorize the dogs as either badgers or hedgehogs because that's all it knows until we give it more information about dogs. Making sure information is formatted so the system can use it and providing the right instructions and data models are essential steps toward deriving useful data insights from AI.
If you're still curious and want to know more about how ADP® uses data and AI in its systems and tools, here are interviews with two of ADP's data science experts:
How Data Becomes Insight: The Right Data Matters — An interview with Marc Rind, VP of Product Development & Chief Data Scientist for ADP Analytics and Big Data, about how to get the right data and make it usable by an AI system.
How Data Becomes Insight: Designing for Insight — An interview with Manoj Oleti, Lead Data Scientist/Technical Architect at ADP, about how the organization develops data insights with the help of AI.