What a furniture maker can teach us about the use of artificial intelligence
There is a well-established master furniture maker in upstate New York named Jim. I know of Jim because he built the kitchen table in my parents’ house. He builds everything from scratch — kitchen tables, bureaus, custom cabinetry and more — all to exacting specifications. Each one of his pieces is just as much a work of art as it is a piece of functional furniture.
A few years ago, I had the opportunity to visit Jim’s workshop and see the environment in which he performs his craft. Located in a barn overlooking a hillside in a rural community, Jim has stockpiled wood and fixtures carefully collected from various regions around the country. His saws, lathes and hand tools are placed throughout his workshop in an arrangement that allows him to conceptualize his ideas and bring them to life.
What I found interesting about Jim’s workshop is that the tools he has are the same tools available to me. I can purchase the exact same equipment at my local hardware store or wood working center. The stock of wood and fixtures he has don’t represent exclusive or otherwise impossible to obtain materials if you know where to look.
In many ways the woodworking tools and materials available to Jim are reflective of the algorithms and data available to all computer scientists. Just as I can purchase the same table saw Jim uses to cut the planks that make his tabletops, any computer scientist can deploy TensorFlow to examine rich datasets for classification.
There is an important difference between Jim and me. I can buy the same tools as Jim. I can obtain the same materials as Jim. If I were so bold, I could even take up residence in his workshop and have the exact same resources Jim uses to build furniture. Even with all of that, I know I can’t possibly replicate what Jim produces. He has 30 years of experience perfecting his craft and is a master woodworker. I, however, am not.
My visit to Jim’s workshop all those years ago was based on the belief that I would see something there that explained what he produces. I thought perhaps he had some special tool, some special piece of hardware, some special stock of wood. That was not the case. The special thing in Jim’s workshop is Jim himself.
When contemplating the use of software, I believe it is critical to think of it as one of many tools deployed to help solve a problem — not a stand-alone solution to a problem itself. Like a master craftsman who uses saws, drills, sanders and lathes to build a fine piece of furniture, so too can a computational scientist leverage artificial intelligence, databases and web services to solve a specific problem. The critical element here is correctly defining the problem and then selecting the right combination and use of computational tools to solve it. As with building any complex object or system, the tools alone cannot create a meaningful result — it’s how those tools are used by the craftsman that yields something of value.