Gen AI in Process Industries: An Interview with Miguel Gonzalez, Chief Procurement Officer at DuPont

Yan: I'm joined by Miguel Gonzalez who works as a Chief Procurement Officer at DuPont. Miguel comes with a wealth of Global Experience having worked all across the world with well-known multinationals. He's been with DuPont since 2015 and became the global Chief Procurement Officer in 2017. 

With all the talk in the world right now about AI and Gen AI, I was hoping that you might be able to clarify for us a little bit more about the difference between AI and Gen AI, and how this is influencing the chemical industry at the moment.

Miguel: I think it's good to clarify the difference between let's start even with ML (Machine Learning), the Bots, the robots that have been around for a long time, and you know a great productivity tool then we have AI (Artificial Intelligence). It has been around for already many years and we all have benefited from AI.

The main difference in basic terms, AI is going to get available information, it's going to extract, it's going to manipulate, it's going to show you data but it's going to be using existing data.

Generative AI, is going to generate non-existing output, it's going to create something new. I think that's a breakthrough because you just get your data, manipulate the data, it really helps but when you know you are creating new things and you know on the personal side when you see some of those tools creating images that they don't exist AI is going to find pictures and images and show you what you're looking for. Generative is going to create something that does not exist, that's probably the main difference.

There are a couple of things that we're starting to see, opportunities like Risk Management. Right now, risk management, and contingency plans, is the big thing for all of us. How do we secure supply after COVID and the geopolitical situation we have seen in the world? Now we can use AI to help us, imagine the tool finding new suppliers finding new routes not only monitoring social media but also recommending actions because right now we have a lot of tools to monitor social media then we find out there is an explosion in one port. Now we need humans to do a lot of finding and actions. 

The other is Data Analytics, AI, what's a game changer in terms of data analytics and shows us data in multiple ways in seconds instead of running all these Excel pivot tables. Gen AI is going to bring us to the next level. Think about SKU rationalisation, not only showing us the data and what is the tail but also finding ways of eliminating the tail, identifying duplicates and alternate suppliers for some of the SKUs we have.

The other thing we are exploring is Network Modelling, not just showing the data where you have your SKUs, where you have inventory, where you have your shipping points, but recommending alternate solutions and finding those warehouses, route carriers, and suppliers to optimise your network. As you can see, we're just starting, we're just scratching the surface. 

Watch the full interview here:

Yan: When we met in the U.S. in Philadelphia earlier in the year, you also mentioned that this is the biggest impact on the industry at least for what you're seeing. Now these are the opportunities, are there any dangers or challenges in working with Gen AI and how might this impact your role in the industry?

Miguel: We all heard at the beginning this is going to be the end of the world. It's okay, people always take things out of context but like everything else that's new, we need to learn, we need to start playing with it. Also we have heard a lot of mistakes and errors and everybody says "you see why it's not working. Look what that lawyer did." And yes that's okay as part of a new thing we're making mistakes but we need to learn from those. One of the things we already learned is most of those errors came down to a lack of human supervision of the outcomes. So do not assume the output is 100% correct, it's not. By default, it's not. It will get better, and better, and better as part of that conversation giving feedback. It's going to get better, but do not just copy-paste the output. That will be a major mistake. The tool is making things up, it's part of the generative creating things but it doesn't know if what it's creating is accurate. It comes to us to give feedback and make sure that the output is something we can use.

Also, the other thing is that the data is limited. The latest version of ChatGPT and most tools that are similar is dating back now to April 2023, so it has limited knowledge of the current world. If you try to build something based on yesterday's news, it doesn't have that information. I believe in the latest versions, now the tools have access to the to the web in real time, so it's going to get better and better but for the last few months and even the Version 3.5, the data was back to earlier 2022. We had examples that the tool didn't even know that there is a war in Ukraine and it was going back to Crimea 2014.

The other most important thing in terms of risk, is you have to protect your data, your IP, everything you load, everything you say, goes into the open. It becomes available for everybody. The first thing you have to do in your company is install the tool behind your firewall, that way you get access to the wall to all the data out there but nothing of what you load or nothing of that you give access to the tool gets out so you're protecting your confidential data, your IP. 

Stay up to date: Sign up here for our global newsletter OBSERVE, and receive the latest news in leadership and top talent, industry insights, and events directly to your inbox.