Computers are changing the way we work. It is probably too broad statement. But if I think about the fact today is Friday afternoon, it should be fine :). I want to take a bit futuristic perspective today. Google, internet and computing are good reason why our everyday habits today are different from what we had 10 years ago. Back in the beginning of 2000s we’ve been buying paper maps before going on vacation and kept paper books with phone numbers of people we need. Look how is it different now. Maybe we still need to make a hotel reservation before the trip, but most of the thing we do can be achievable online via internet and mobile devices.
A month ago, I posted about connecting digital and physical entities. I was inspired by Jeff Kowalski presentation at AU 2014. You can get a transcript and video by navigating to the following link. The idea of machine learning and “training” computer brain to find an optimal design is inspiring. The following passage from Kowalski’s presentation is a key in my view:
…we’re working on ways to better understand and navigate existing solutions that might be relevant to your next design project. Using machine learning algorithms, we can now discover patterns inherent in huge collections of millions of 3D models. In short, we can now discover and expose the content and context of all the current designs, for all the next designs. Taxonomies are based on organizing things with shared characteristics. But they don’t really concern themselves with the relationships those things have with other types of things — something we could call context. Adding context reveals not only what things are, but also expresses what they’re for, what they do, and how they work.
Nature explores all of the solutions that optimize performance for a given environment — what we call evolution. We need to do the same thing with our designs. But first we have to stop “telling the computer what to do,” and instead, start “telling the computer what we want to achieve.” With Generative Design, by giving the computer a set of parameters that express your overall goals, the system will use algorithms to explore all of the best possible permutations of a solution through successive generations, until the best one is found.
Another time, I’ve was recently thinking about artificial intelligence, machine learning and self-organized systems was my article – How PLM can build itself using AI technologies. The idea of The Grid that allows to self organize website based on a set of input parameters and content learning is interesting. It made me think about future PLM system that self-define system behaviors based on the capturing of information and processes from a manufacturing company.
The article Google search will be your brain put another interesting perspective on the evolution of computer and information system. Take some time over the weekend and read the article. The story of neural nets is fascinating and if you think about a potential to train the net with the knowledge of design, it can help to capture requirements and design commands in the future. Here is an interesting passage explaining how neural nets are working from the article:
Neural nets are modeled on the way biological brains learn. When you attempt a new task, a certain set of neurons will fire. You observe the results, and in subsequent trials your brain uses feedback to adjust which neurons get activated. Over time, the connections between some pairs of neurons grow stronger and other links weaken, laying the foundation of a memory.
A neural net essentially replicates this process in code. But instead of duplicating the dazzlingly complex tangle of neurons in a human brain, a neural net, which is much smaller, has its neurons organized neatly into layers. In the first layer (or first few layers) are feature detectors, a computational version of the human senses. When a computer feeds input into a neural net—say, a database of images, sounds or text files—the system learns what those files are by detecting the presence or absence of what it determines as key features in them.
So, who knows… maybe in a not very far future CAD and PLM systems will be providing a specific search based experience helping engineers to design and manufacturing in a completely different way.
What is my conclusion? While it still sounds like a dream, I can see some potential in making design work looks similar to search for an optimal solution with specific constraints and parameters. A well trained algorithm can do the work in the future. Just thinking about that can fire so many questions – how long will take to train the net, what will be a role of engineers in the future design and many others. But these are just my thoughts… Maybe it will inspire you too. Have a great weekend!
Best, Oleg
