Artificial intelligence and machine learning are going through the renaissance period these days. Few days ago, I shared my some of my thoughts here – Your next PLM workflow manager will be … bot. Actually things are getting real these days. Imagine the following conversation between engineer or manager in manufacturing organization and smartphone in a not very distant future:
Bot: You have an email about issue with airbag model XYZ-100
Engineer: Please assign high priority and initiate change request.
Bot: I just discovered it can impact AB-1 and AB-2 car models.
Bot: Do you want to initiate engineering change request?
Engineer: Yes and please organize a meeting to review field samples
Engineer: Also, please invite extended team and search for similar reports for the last 2 years
Bot: You will have it done until tomorrow
Manager: Are there pending ECOs waiting more than 2 days?
Bot: You have 2 unapproved ECOs. Would you like to proceed with approval
Manager: Yes, please.
Bot: Here is the link to a playcast with information about ECO-101. Pending your approval.
Bot: I discovered one more ECO-1022 is pending approval of electronic engineer
Manager: Please activate video call and connect me in the next 6 hours.
Bot: Will be done.
Do you think I’m crazy? Of course, there is a possibility I’m daydreaming. However, if you follow news, you can see that some companies are already working on bots that can automatically answer your emails and serve as you personal assistant in online chat applications.
Google blog article – Computer, respond to this email speaks about experimental development of a feature called “Smart Reply” in Gmail. The following passage from Google blog can give you a general idea how it works.
A naive attempt to build a response generation system might depend on hand-crafted rules for common reply scenarios. But in practice, any engineer’s ability to invent “rules” would be quickly outstripped by the tremendous diversity with which real people communicate. A machine-learned system, by contrast, implicitly captures diverse situations, writing styles, and tones. These systems generalize better, and handle completely new inputs more gracefully than brittle, rule-based systems ever could.
If you want more details, the following publication – Neural Conversation Model by Google engineers is explaining how they use neural networks to map sequences to sequences. The model is based on a recurrent neural network which reads the input sequence one token at a time, and predicts theoutput sequence, also one token at a time.
My second example is coming from Facebook M assistant for messenger. Read more in the following Re/Code article.
That’s the premise behind “M,” a new “personal digital assistant” that lives within Facebook’s standalone messaging app, Messenger. Unlike Apple’s Siri or Google Now, you’ll primarily interact with M through text (although you can send M a voice recording). The assistant is powered by artificial intelligence, the advanced technology Facebook is developing to help its products think and act more like humans.
Right now, Facebook is training M with supervised learning, a process where the computer learns by example from what human trainers teach it. If a user asks A, you respond B. Eventually, the idea is that M will know enough to operate without a human handler. Facebook has a team of people building neural networks — applications that help machines think and act like humans — and many of those applications are already live inside of M, Schroepfer says.
What is my conclusion? I think, I’m not completely lost my mind. Today, PLM vendors are trying to bring traditional PLM organizational paradigms of data management and workflows into a new world of product complexity, disparate data and mobility. Workflow is just one example, but I like it. Most of workflows are complex and expensive to maintain. On the other side, organizations are complex, global and influx. It is a time to think about new paradigm. Just my thoughts…