A WORD ABOUT ARTIFICIAL INTELLIGENCE
FEAR
ABOUT AI, REALLY?
Someone s
dystopian future may have all design and production jobs done by Artificial
Intelligence machines. The first problem with that scenario is that AI machines
are not potential customers like employees. Eliminating jobs throughout society,
would mean eliminating customers throughout society. A second problem is
predictable, competitive AI models will likely develop differing results in
favor of their respective sponsors. That sets up a modern tower of Babble. New employees will need to sift through AI
results, and as needed, fix what AI does. AI will affect employment, especially
in the chaos created by lack of understanding AI. The workforce, the people,
will adapt and people will still be needed.
ASSUMPTION
INTELLIGENCE
Artificial
Intelligence (AI) has been around for decades, but has become a more active
topic recently. I view AI as Assumption Intelligence , meaning it is a computer
program written to draw from data libraries and sensors to determine a solution
to a particular need or problem. The assumption is the solving strategy, built
into the process created by the software developer. Their difficult job is to
consider a large variety of problems, and then write a single program that comes
up with an individualized solution for each. The result is a working solution . A working solution will do the job, but it is
not the best solution for any specific problem, because a custom-tailored
solution can do better. In a lean
manufacture mind set, there is always an improvement to make.
For
Example: In production, Computer Aided Machining (CAM) software is used in the
creation of cutting programs for components of all shapes and sizes. The
assumptions built into the program s code will/should come up with valid tool
paths (g-code) to machine the component. But, a
solution pathway developed to process parts of all shapes and sizes, will
likely not result in the optimum g-code for any one specific part. An engineer
or machinist familiar with g-code, can amend lines in the program and make
improvements. Potential, cost-saving,
improvements are often readily visible in the initial run of the AI s program.
Watching the program often begs the question: why is it doing that? It is
doing that because of the generalized assumptions built into the CAM
software. Let AI quickly write the 10,000 (or so) line program, then take what
AI made and improve it. Those improvements can save considerable time and money
in production.
Good AI
is helpful, but it needs to be fed a start and then it needs a finish. If you
think of the task as a flight of stairs, AI can assist all the middle steps,
but the first steps and the last steps are yours to make. You have to define
the problem up front in order to design, and later, you must evaluate and
improve the production process. Good AI does not eliminate the worker. It
is a step saver and can improve the result and lower risks.
AI IN
PRODUCT DEVELOPMENT
If you
purchase an AI program and ask it to come up with a brand new
product idea (with no further input from you), will its answer belong to you or
the company that created the AI software?
Whose idea would it be?
Like the
flight of steps mentioned above, product development s first steps are a set of
requirements for what is to be designed and made (product or process). Requirements define the needed safety,
geometry, purpose, and performance, of the goal product . Call those first
steps preliminary design or logically sketching out the product. This step
documents the idea and allows ownership of that idea to be established
(provisional patent). From that point forward the process progresses up the
stairs to: drafting (detailed design); prototyping + testing, first articles + testing;
and full production startup. The higher
the step in the process, the more money is laid out to accomplish it. Money is
risk. The last step, ramping up for production, is the largest cost before a
product can start to return investment. Computers assist with those latter
steps now in design and manufacture. AI
has been and will be a beneficial factor, especially to lower risks and speed
the development process. People, though, will still need to define the idea,
evaluate the testing results and determine if the product is feasible before
proceeding to pay for the next step.
AI helps
product development with features built into software like CAD and CAM
software. Features guide designers to
properly define their models, evaluate fit, virtually test strength and perform
other simulations. All of these eliminate potential errors in the lower cost
design phases (steps) and speed up development time. AI can eliminate risks
within the steps of a development process. It cannot do the process, but it can
significantly help.