Wednesday, June 12, 2024

How Automation and Analytics Work Together

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

How can automation and analytics drive digital transformation?

As the CTO of GE Digital, Colin Parris – formerly a VP at IBM for a decade – is actively involved with shaping how automation and analytics drive business productivity and digital transformation. This includes: 

  • How businesses increase output at lower costs.
  • How lean business transformation helps customers get to value faster.
  • How technology can support customers in light of COVID-19.

Please join this wide-ranging discussion with Colin Parris, a key thought leader in transforming business.

Download the podcast:


How has COVID changed how industrial companies are thinking about digital transformation?

  • “I think we’re seeing two things change. One is this notion of acceleration. These firms are seeing that they need to accelerate it and they can. And the second is this thing I would call normalization. [Digital transformation] is now becoming part of the fabric and accepted.”
  • “I remember having a discussion with the CEO of NYPA in which he was saying he had discussed with his team, ‘How long will it take for us to go and do remote operations?’ They said that transformation would take three to four, maybe five years. They did it in five days because of COVID, because they had to.”
  • “With any kind of digital transformation, companies are looking for two things. They were looking for productivity, and they were looking to manage the market dynamics. In terms of productivity, they always have the view that software and automation can actually give them more capability.”
  • “The second thing they’re saying is that in terms of handling market dynamics, they always knew they have to plan long term. Now, with COVID, you have to think short term. You have to think doing it right now, right away.”
  • “One other quick example, Haverhill, a water utility in Massachusetts. We have a control system working with them. We’ve been working with them for years on actually remote monitoring. When you tie that to the automation system, they control something like 5 million gallons of water annually. They were able to move from a control system where you can have one person off-site running this control system for this entire 58-59 thousand homes. Again, COVID is actually proving these things out. A powerful way to accelerate it and make technology the new normal.”

What is a digital twin, in terms of digital transformation? 

  • “We, as engineers, are custom-building models. We build a model, and we use the model to design something. A model is a way you think something works. And then we use that to design, we build a model to understand what the problem looks like, and we use that to figure out problems. We all use models.”
  • “What a twin is, it’s a model, but it has two different characteristics. It’s a living learning model. In other words, after I finish to use the model and design, I throw it away because I built the thing. How about if you keep that model alive or parts of it and then constantly feed it real-time data so the model changes to reflect the asset as it’s being used right now in the world? So it’s a living model. We have something like 1.2 million of these twins built.”
  • “Let me use a utility example, CPV, Competitive Power Ventures. And what they are using their twin to do, first of all, is to predict. Then they use the same twin to optimize.”
  • “So now we have something, a twin controlling how high can I turn up the temperature so I could heat it up to get more electricity while not damaging the asset? So a twin, we have one example of doing both, prediction and optimization.”
  • “Initially, to form the digital twin, you need the expertise. But once you have a twin, the model, you can deploy the model and then that model then communicates with regular humans who use it.”
  • “An engineer will say, ‘Well, what does the twin think will fail?’ And if the twin says, ‘Well, I think these four parts will fail,’ engineer can say, ‘You know what, that makes sense.’ And they’ll order those four parts way in advance. This way, you don’t tie up your money ordering parts that you don’t need or you don’t have the wrong parts when it comes.”

How will artificial intelligence will play a role in digital twin deployments in the future?

  • “So I’ve found a way to deliver value through the twins. I can tell you how to save money, or a how to gain revenue in a variety of ways. Now the problem is, every time we build these things, how do you actually deploy it?”
  • “You’ll say, ‘What about business risk?’ “Collin, I don’t trust this model. This is a model, a neuro network, a black box. I don’t trust it.” How should I think about that? Okay, so what we began working on a couple of years ago with something called Humble AI.”
  • “So what Humble AI does is that it knows what is the zone in which I’m competent. Ah, seven to eight I’m competent, that’s my zone of competency. Outside the zone of competency, I use the regular models you had before. Then I ask for help. Then I say, ‘Can I get more data about, data in the 12 mile per hour region?’ So this is Humble AI.”
  • “So what we have now is we have an AI system that can take that information before it gets there. We can get to that information off the turbine, and create the graph. And say, ‘Well, given that graph and comparing it to what the designers have written in the manual, this is a pitch problem.’ That person then will ask, ‘Why do I believe that?’ You pull out of the manual, ‘Here’s the exact graph I compared it against to know why it was exactly going to be a pitch problem.’ And so now the human says ‘Alright, that is okay, I have seen this in my manual before, I think it’s right.’ This is Explainable AI.”
  • “What if two wind turbines could talk or the wind turbines in one big farm could talk to each other and one wind turbine could say, ‘I am actually feeling this funny vibration, has any of you ever felt it before?’ And one of them could say, “I felt it before, this was the problem then.”
  • “You can have the wind turbines communicating and answering their own questions and then telling it to us. So we have been developing something called an Emergent Language. It’s a new form of AI in which we are looking at the way data comes off and getting wind turbines to talk to each other, to have a language in which they can ask each other questions and we’ve had some amazing results.”

If the wind turbines are connected in that way, using AI, will people lose their jobs, or will it actually create jobs?

  • “Oh, no. There are two things we do it for. One we do it for safety. Because if there are things that are about to happen, we have challenges.”
  • “The second thing is that if we begin to do that, we begin to find ways now of getting the human involved at the right time. So imagine two wind turbines are talking, one gets an idea, ‘After I’ve spent $600 million, am I gonna trust that alone?’ No. The human then becomes, ‘I’m gonna be the one who advocates and who understands, and who works through it.’ And so this is where the humans get involved.”
  • “The next thing that will happen is that nothing stays the same. There gonna be other new problems that show up, I can have the humans focusing on the new problems. What I don’t want is my best people focusing on problems we’ve solved, but you just didn’t know you solved them before.”
  • “So I don’t see loss of jobs, especially as there are more and more wind farms that are growing. I see humans working at what they do best and not having to do that work in which it’s dirty or dangerous or dull.”

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles