Automation: When Should We Stop?

When it comes to automation, things are just getting started.


Automating tasks has become quite popular. Driver assist is a good example. So is any machine learning application. You may have noticed that you can now get deliveries in less than 24 hours on Amazon. There is huge automation behind that service. So the question becomes, when is enough, enough? When does automation hit the law of diminishing return? When does this all become “creeping featurism” vs. real impact?

Figure 1

The short answer is “never.” We are embarking on a series of life-changing events all around us that are rooted in deep, sophisticated automation. Where it stops, who knows? One thing is clear – things are just getting started.

Let me describe an experience we’ve had related to automating access to semiconductor manufacturing technology. Over the past year, we’ve found that our online quoting tools (MPW Explorer in particular) have resonated very well with the academic research community. These folks need to prototype their research, typically using multi-project wafer technology, to demonstrate real results. The folks in this community reacted very strongly to the transparent, always-on, easy-to-explore experience.

At first, everyone raved about how quick and easy it was to get an accurate quotation for an MPW run. Then we started to book business as a result. And then things changed. That same community started giving us different feedback. “The quotation process was great, but now that I’ve signed the quote, what happened to the automation?” PDF documents to sign. Complex questionnaires to fill out. Lots of emails to nail down option choices. FTP sites to battle with in order to upload GDSII files. You get the picture.

This is a perfect example of too little automation. We’ve since corrected the problem. Signatures are gone from the process. Questionnaires have been replaced with online forms that are intuitive and easy. All process option choices are specified this way, too. GDSII files are uploaded with a simple point and click. All of this comprises the project management function that is now in beta test as part of MPW Explorer. Reception has been quite positive. There’s just never enough automation it seems. Our job is far from done – there are many more ways to simplify access to semiconductor technology. It’s all part of our roadmap, and there really is no end in sight.

Figure 2