Designing and building a product is comprised of a set of tasks. There are 3 places where the knowledge about each task can live:
- people's heads
When knowledge lives in people's heads we have what we call bespoke work. There is usually one creator and they understand every aspect of the design process, and know why every decision was made.
Knowledge lives in processes when people always solve a problem the same way, to the point where they no longer know exactly why they do it. All doctors wash their hands after touching a patient, but they aren't actually considering that they might spread an infection if that patient was sick - they just always wash their hands. The solution to that particular problem is no longer in the doctor's head, it's in the process they apply. Most modern products rely on processes to be made, since there is so much knowledge involved that no one could understand it all.
Lastly, technology allows us to grab processes and automate them. Think of light sensors - you used to have a process, when it was dark, you flipped a switch and the lights went on. You didn't actively think about the problem you were solving, but you still had to do it. With the light sensor, the lights go on when you need them to go on. You neither think about nor do you do anything. The problem is solved, for as long as the machine works.
Technological progress is in itself a process of moving knowledge from people's heads, to processes and then to machines. When you move knowledge to a process, suddenly less qualified people can do the same work; when you move knowledge to a machine, no one does that work anymore. This is how productivity grows and how we become free to be creative and solve bigger, more interesting problems and raise everyone's quality of life.
Our current CAD tools still keep a lot of knowledge in the designer's head, or in processes they might apply. Knowing how different materials interact with each other, and design considerations you must have when working with a certain material is something you always have to hold in your head: the software won't prevent you from bending concrete or casting wood. Adding thermal insulation to a wall is something you actively decide - the software will give you no feedback if you don't. You leave enough space around furniture because you follow a process - if you think this is obvious, just look at an architecture student's first project (sense of scale is a learned skill).
These are all important things, but do we want to be thinking about them when we design, or do we want to be thinking about how to create great atmospheres and products people love?
Processes increase productivity, and give us more time to consider the essence of what we're creating, but they're not without risks. For one, processes become mindless tasks you repeat everytime. Neither is that a good use of our time, nor are we particularly good at repeating tasks consistently for long periods of time. At the same time, processes are solutions to problems we're not questioning anymore, which stiffles innovation.
Why do we need to move more knowledge to machines
Automation is helpful to become even more productive and eliminate repetitive work, but also to make innovation cheaper. You see, even if you somehow get to question a process, the cost of experimenting with an alternative is to execute your new idea, using whatever time that takes, and if it fails, taking the time to execute the initial process anyway. When a task is automated, you only pay for your experiment, since the original solution is applied by itself anyway. Also, when knowledge lives in a process, proposing any other solution often sounds ridiculous, because "we've always done it that way". Yet, history tends to show us that the processes we apply are often the ridiculous ones.
Processes are an improvement from bespoke work, but they are still a time sink. While you're no longer solving the problem from scratch, you still have to hold the process in your head. This costs you not only time, but focus in your creative work. By automating a process, you increase your cognitive capacity for lateral thinking and considering new solutions.
People will always be solving some problem in their heads. By moving knowledge to machines, we can make sure they are new problems, and not always the same ones.