Transforming R&D and IP Strategy with Ashish Bodhankar

Erin Pearson: Hey everyone! Welcome to today’s episode of Decisions Now. I’m your host, Erin Pearson, and I’m very excited because today we have Ashish Bodhankar, who’s the IP Strategist Leader at Dow Chemical Company, joining us today. Welcome, Ashish, how are you?

Ashish Bodhankar: I’m doing very well, Erin. Thank you so much for having me. Excited for having this dialogue with you about AI.

EP: Yeah, no, I think it’ll be great. It’s a topic that we haven’t explored too much on the podcast yet. So, it’s about intellectual property and the use of AI within it. So, just to kick us off and give a little bit of background, what inspired you to have a career in intellectual property management? And how have you been seeing the role evolve over the years?

AB: Yeah, to be honest, it was not a first choice, correct? But I was a chemical engineer as I was always excited about technology and technology management, how things get commercialized. And I mean, my first job was with Evalueserve. So, Evalueserve happened by accident. And I got a chance to work on a different set of inventions coming from different companies, which eventually got me excited that I was reading through new technologies, how companies work, how do they build their intellectual capital, and how important it is actually for the companies. So, my interest started going up, I looked at the case laws, looked at how licensing works, how people are making money out of this. So, it kind of, my interest and excitement kind of kept going up. Then when I joined Dow, I also saw the other side of the story as to how exactly the technology is put to use, how intellectual property is actually utilized, why is it important, how it’s a main part of the strategy, and how it’s a key consideration at the board. All this kept me going, and it’s been close to 20 years in this field. I’m pretty happy and very satisfied so far.

EP: Yeah, I think that’s cool. I think you had a unique opportunity that you got to start kind of in a consulting-oriented role where you got to see how a lot of other companies approach the IP area and how they brought it into their business. And now, you’re on the other side of it and really applying the work within your role as well. So, I think that’s just a really cool perspective that you’re able to bring to the table as well.

AB: Absolutely.

EP: So, within that, especially when we look at the past few years, as with any industry, but there’s been a ton of development around artificial intelligence. So how has that been impacting the R&D strategy that you have at Dow Chemical? And how can businesses measure the ROI when they are starting to implement AI into their program?

AB: Thanks for the question. I think AI is a buzzword today, as we see many things floating around, and everybody’s talking about ChatGPT, predictive AI, generative AI, and things like that. But if I step back a little, I think this is not new in a sense, this has always been a work in progress. If you go back 10 years, then we started seeing things like modeling in the chemical world, as well as in general, predictive modelings were used, statistical techniques were being used to predict and analyze information. We slowly have things like natural language processing, text-based analysis, you have text cluster analysis, you have these document identifiers and then how do you compare the two documents if you look at Microsoft Word, how do they compare to documents and things like that. So, I think these things existed, but what has happened right now is the technology has really picked up pace right now and then AI has enhanced itself substantially. That’s what I would say. And it does have an impact on R&D, in general, the day-to-day functioning of R&D, but at the same time, how do we look at it more strategically? How can this give us a competitive advantage in the field that we play? Of course, material science remains material science, but then how can AI help us drive things? So, it certainly is helping business. There are multiple ways by which AI can help R&D strategy. We can look at it to speed up R&D. For example, can we quickly analyze formulations? Can we quickly analyze data that we have from experiments? Can we look at forecasting of what kind of formulation would work for a given property or end application. So, all these things, and we are implementing some of these things today. So, a lot of these things are works in progress. So primarily from an R&D perspective, it can help us identify new opportunities. It can also speed up the time that is taken for R&D. Typically, in the R&D chemical world you will see it is going to take a lot of time to develop a new product, something physical, and then bring it to the market. So, I think that time frame would be substantially reduced with the work that is going on in AI, both generative as well as predictive AI. It is helping us analyze complex patterns in the data experiments. We can use the existing data to see what other formulations can happen. And part of this, again, as I said, this was this is not new. It existed, like high throughput experimentation technique is probably the starting point, which was there in early 2000s. So, it’s progressing…the speed with which we’re progressing is probably now faster.

EP: No, absolutely. I think with a lot of AI, I think a lot of the use cases, they end up coming down to, it helps with speed and it helps with predictions. So, I mean, you mentioned a lot of things, but it just has this ability to analyze so much data at once that it’s just naturally going to speed up that process. And then, whether it’s sorting through trends and just pulling them out and doing a trend analysis or prediction on what’s going to happen or identifying new opportunities, it’s all just based on predictions of what this data set that you have holds. So, I think those are great areas that can have a huge impact on R&D, especially because so much money is invested in it on an annual basis, that having better data, faster data to make better decisions or go to market faster or go to market with a higher level of prediction will really help you in the long run as well.

AB: Yeah, and on a day-to-day basis, just to add to that, on a day-to-day basis, it helps us optimize experimental conditions, correct? So rather than doing thousands of experiments, I mean, high throughput experimentation enabled the R&D person to do thousands of experiments in a shorter duration of time. Now imagine if we are able to kind of improve the accuracy of that, then that would further save us time, correct? I don’t need to sit down and think about different experiments to do. So, the design of experiment part has really improved using some of these techniques.

EP: Yeah, that makes sense. But one of the things that’s happening is a lot of people are still…there’s a lot of struggles with just people looking to implement this or getting adoption within the organization. So, what are some of the biggest challenges that you’re facing when you’re trying to deploy AI into the R&D process that you have? And what are the methods that you’re using to overcome these strategies?

AB: One of the key challenges is right now in understanding that this is applicable to our industry. So, when you look at chemical and material science, people think, what has AI got to do with it? I mean, we’re going to make materials, correct? So how would a computer make materials? How is this going to be used? So, that is one challenge in terms of accepting that this can play a role in the organization or in R&D in particular. So, we need to understand what these systems are and how they function really well. So, as a chemist, I’m not equipped today to really understand how these systems work, because that’s not my training, correct? I’m trained more to do with the lab. How do I conduct experiments? I’m not trained in terms of how these things are going to help me have an impact. I think that’s another challenge that chemists would have, probably, at this point in time. And the other challenge is, we need to make sure that the current systems of AI or machine learning, whatever are in place, are more accurate and reliable. So, one of the challenges is to ensure that when we are using such kind of a system or we are developing something, then they are giving us reliable and accurate results. Otherwise, it’s going to be a waste of time and things like that. And the other element is we need to be mindful of all the data privacy issues, the security concerns, etc. So, I think those are some of the challenges. Probably the biggest one is, right now, the AI systems are purely based on what data is keyed in for them. So, your systems are as good as what you key in. I mean, we had a saying, whatever you sow in, that’s what you will reap. So exactly that’s what AI is giving you right now. But having the right talents, making sure that you are educating your R&D folks, and then you’re making sure that the quality of your data is keyed in well. You also are following the right security, safety, and ethics policies around managing this kind of information. I think I see them to be bigger challenges, specifically in an industry like chemicals, to adopt to this new technology. And maybe one more point is, it’s a matter of how exactly this is going to play. Right now, we’re not able to visualize how exactly this is going to play a role, correct? We see some predictive analysis. We see ChatGPT giving out, writing out essays, and giving us information from the public domain and things. But we don’t really see how this is going to impact the actual R&D. So right now, the challenge that I see is, even I’m not able to fully visualize how it’s exactly going to play. The automotive industry in the past probably would not have visualized that automotive electronics is actually the key now, forget about the basic automotive functioning. And suddenly they’re working with the Googles, the Samsungs, the Qualcomms who are actually writing the codes for their automotives. 10 years back or 15 years back, probably that was not the case. So, we just don’t know how it would play out in our industry in particular.

EP: Well, I think that’s interesting. And one of the points that you touched on in there is it also seems like, I mean, the amount of importance that it can’t be understated that the data that goes in is going to significantly impact the predictions or the answers that it’s giving you on the other end. So, you really have to be very upfront and clear about what business question are you really trying to solve for to understand what’s the right data set that I need in the first place. And then when people are using this, it essentially needs to be stated what is that source of data that was used in the first place. A lot of statistics that get used over the past 50 years or whenever, but things get cited as if they’re a more general population, when in reality the study was far more specific, and then people just use them because they’re big wow stats instead.

AB: Absolutely. Same principles apply. I think if we just survey 10 people for a country like India, let’s say, the population being 1.4 billion. 10 people saying yes to a particular thing doesn’t necessarily reciprocate with India’s opinion. So, the same principles would apply when we use these systems.

EP: Yeah, so I think that level of visibility will hopefully be coming out. But one of the things within that and that you were also touching on is, and I think that relates to this as well, is how can businesses ensure that their AI systems are ethically sound and also aligned with the values that the company is coming out with to make sure that, whether it’s the data that comes in is the correct type of data or the decisions that are being made on it are ethical and transparent as well. So, how do you see that coming together?

AB: Yeah, I think as we were discussing, we certainly need to, as a business or as a company, we certainly need to make sure that we have guidelines or policy around the use of these systems, which would ensure that there is a level of transparency in terms of how exactly we’re gathering the data and then how are we going to use this, correct? And that principle is probably existing today, correct? It applies to the same survey that is conducted by news agencies or as you said, correct scientific reports being published about a certain thing. So many reports are there on diet and nutrition, correct? So, we need to be mindful of how exactly the survey was done, who was involved in it, how the data got collected, and how are we planning to use it. So as a company and also as a business element, we need to have that policy in place, it’s just like safety policies. What kind of safety nomenclature you have in your company, how are you implementing those policies, how are you doing on that front? And safety comes under ethics in a way, correct? And I would apply the same principles for AI systems when they get deployed and when we actually share results with the customers also. It’s very, very important.

EP: Yeah, I totally agree there. I think having that contextualization…and it needs to be done in an executive summary style format so people don’t ignore it either. You can’t create these hundreds-some page documents that it describes all of the research practices and all the detail behind it, because no one’s going to read into that. But you just need a very one-sentence answer in terms of what’s the context behind this data and how did we get there.

AB: Correct. And then, of course, the other element is the economics behind, correct? I mean, if I have a lot of money and I have access to these systems, I am privileged, correct? And I can do things which probably companies or individuals who do not have the money power to have these systems in place. So, there’s so much imbalance that these things can create, not just from a business perspective, but also from a social angle. So, I think strong guidelines would be needed for sure.

EP: Yeah, that makes sense. So then what advice would you have for businesses that are really just starting to deploy AI in their strategy, or specifically in their IP strategy?

AB: I think like everything, you should start small, whether you’re doing a diet or you’re trying to work on your health, you’re not going to run the marathon directly, correct? So, you’re going to start small and focus small. I think have smaller projects and assignments and then kind of try those things out to make sure that you understand the nuances, you understand how exactly the systems work and then also evaluate what kind of benefits are you actually getting from doing this kind of work. At the same time, you also need to make sure that you have the right set of people. As I said before, as a chemist, I just don’t know how exactly these things work, or I’m not the best user. So, you need to make sure that you have the right collaborations for your test cases as well. And you have the people who know the systems, who know how things work. Like, can I have someone who would guide me, someone who would be able to help me manage the learning curve piece? I think those are the things that can be done by businesses to begin with. Then there are certain things which, without compromising some of the confidential information or some of the critical R&D products, one can use things like, if I look at it from an IP perspective, then we can always look at doing a general pattern landscape analysis, correct? And then using these systems, how is the system analyzing the data? How is the system using different types of technical information to classify the information? And it’s all public domain knowledge. You can see the impact of using such kind of a system. You can also, I mean, Google has fine relevant pieces of prior art. So, you have a button which kind of clicks fine prior art. And nowadays, new tools have different types of systems like that. And those are good case examples for companies to look at, not just from an IP perspective, but even from an R&D angle. Let’s look at this because this is less harmful data. We have a lot of data which is available. The data is high quality. When you look at patents, for example, they’re granted, approved, examined. The examples are actually conducted. So, it’s all validated information, government data in a way. And then that could be a good test example. But again, start small, have someone to guide you, and you should be able to then understand the systems much better.

EP: Yeah, that makes sense. You want to have quick wins as well to make sure, a)  you’re getting it right, you’re answering the right questions, but you also don’t bite off more than you can chew. So that leads into the next question. So, it’s very important to keep an ongoing evaluation of how things are going and create adaptions based on what is going well? How can I improve that? What’s not going well? How can I reduce those errors? So how would you advise people to create that feedback loop into their program?

AB: I think as we started off, the pace has kind of substantially picked up in terms of the development of this technology. So, it’s very, very critical for us to have that ongoing evaluation and adaptation given things are changing rapidly. And we know for sure that this is going to hit us. So, we cannot say that we are a material science company, and this is not going to hit us or we’re an oil and gas company, we just dig out oil and we sell it. How does this impact us? So, I think it’s going to hit everyone. So, we definitely need to be…leadership needs to be more agile and adaptable. And then that kind of message should flow down, that we should have a digital strategy in place. I would call it a broader umbrella of digital strategy. AI would be a part of it. So, I think a digital strategy team would be a must for an organization who then can do regular evaluations and have those small projects and things like that for the organization so that we are the early adapters, we don’t miss the boat. And then at the same time, we are careful so that it doesn’t necessarily negatively impact us. We kind of stay relevant, stay reliable, stay more accurate. It’s all positive. So, one should take it in that sense. But a digital strategy team is what I would recommend to have, where people from business, as well as R&D and other aspects of business, are also there along with the experts. So, I think that’s the key to have.

EP: Yeah, I think it’s brilliant. And one of the key things that I think you very accurately said is, it’s important no matter what industry you’re in to not think that you’re going to be a laggard or you’re just like, oh, maybe my industry is just like, this is the tried-and-true method. This is what we’ve been doing for forever. It’s going to keep going. Plenty of stories about how companies have done that, and it has not worked in their favor. So, I think, too, what we’ve…I believe what you were saying earlier was a little bit, we’re still in an exploratory area to figure out what are those true groundbreaking use cases. So right now, it’s really about productivity and improving that within the employees and getting adoption, getting buy-in. But over the next couple of years, there’s going to be some really big shifts in terms of how businesses truly operate as they do bring this in. So, when you talk about something like the business strategy team that you just mentioned, there’s obviously a collaboration that’s needed amongst many teams. You have a lot of different skill sets from understanding the business, understanding the domain. You need to be able to take data, turn it into a story. You need the data engineers to be able to actually build the AI. So that’s a lot of collaboration. There’s a lot of different personality types that come in from the people that fill these roles as well. So, how can you foster a culture of collaboration within an organization and really bring these people together and have them collaborate effectively?

AB: Yeah, I mean, collaboration is key, irrespective of whether we’re talking about digitalization or if we want to do the business in a heritage way as well. So, collaboration between functions is certainly essential, whether it’s R&D, commercial, logistics, supply chain, whatever it is. And specifically when you look at things like AI or machine learning kind of things which are coming up right now, I think this becomes all the more important. That’s why I said we probably need to have a digital strategy team wherein we have people from different functions within the organization along with the digital experts so that everybody kind of understands how this is going to impact and how, as a company, we need to deploy this for our success. The ultimate goal is a common goal. We want our company to succeed. We want higher REITs, we want to make more money, and, at the same time, we also need to make sure that we benefit the society in general. So, with that in mind, I think a cross-business team, a cross-functional team is certainly critical. And if you have the buy-in at the leadership level, then that is something which is easy to percolate down. So, I think companies should have those kind of cross-business teams. I mean, not just for AI, but for example, even for IP evaluation, if I’m making decisions on my own, whether I want to file a patent or not, then that’s not necessarily the right decision. I need to have my lawyer in place. I need to have a searcher in place who’s going to give me the right pieces of prior art. The lawyer is going to analyze it. I need to have an R&D opinion as well to distinguish it technically or create an example in the lab. And then finally, I need a businessperson who will tell me, yeah, it makes sense because this is the business that we want to protect. This is how my competition is operating. So, it’s an amalgamation of all these points and not my individual decision because I think it’s patentable subject matter. It has to come in from all these people. So, I think it’s important, and it is essential for AI, specifically because we don’t know what’s going to happen.

EP: Yeah, I mean, it’s an exciting time to be in business, too, because there is so much change that’s happening. So, these types of events don’t actually happen that often to the scale that it is. And you know that you’re in one at the moment too. So, it is very exciting. So, do you have any ideas on common misconceptions about AI being deployed in R&D strategy, and how would you advise to help address them?

AB: One of the common concerns that we hear is AI is going to take up our jobs, is there a cross-function, not just R&D, correct? People feel that across the domains. I would say maybe it’s a valid concern because we just don’t know how things are going to play out, and it’s been hyped in a way that this is completely different than what we had in the past because people often cite the counter-argument saying that we had the Industrial Revolution, that we had the advent of computers, but they never took away the jobs of humans. So, I think probably we need to look at it positively, as you were also saying. I mean, we need to look at how it impacts us in a positive way, how we can use this. Ultimately, it’s a tool which we have developed, and this can help us enhance our own capabilities. This can help us enhance our decision-making capability. This can enhance the profitability of businesses. This can probably help us also on a social front, bring the balance faster maybe. Today we see that society is imbalanced. It can probably bring it faster there on the level of balance. At the same time, the other misconception is, if AI comes in, we just don’t need to do anything. The computer will write. Today, you see ChatGPT writing my daughter’s essay. A rainy day, just write an essay. So, the point being, that’s not necessarily the solution for all problems, correct? Or the initial solution given by an AI system may not be the most…the best one, correct? But it’s a good starting point. It is saving you a lot of effort, energy, and it’s a good quality output, which you can then develop further, which means it gives me a chance to enhance my skills further. So, if I look at it that way, then that’s how we need to look at it. And I think it’s inevitable. So, the sooner we accept the reality that this exists, and as I said, this existed in the past as well, we are progressing. And it’s progress over the last 20 years, 25 years, and it will keep progressing. So, I think that’s what we need to have that change and more openness to adopt this. And I don’t think it’s going to take away the job. Maybe some routine jobs would get altered, but that means it’s an opportunity to develop oneself.

EP: Yeah, well, I think that’s true. It’s less about taking away the jobs and more, some jobs might evolve, or they actually might become more interesting, because I mean, no one really wants to sit there and just do something like super manual all day, because that’s pretty mind-numbing. So, I think it actually creates a new opportunity for people to do things that are much bigger as well. And one thing that you touched on as well is AI or like ChatGPT as an example can be used as a teacher as well. So, I think there’s tons of applications for people who, whether they feel like they’re not getting the skill development that they need, whether it’s they’re in high school and they’re trying to learn a new math concept and maybe they just aren’t on the same page with a teacher or they’re in business and they’re looking for, how do I do a better interview? How do I learn this skill better? Whatever it is, it can help you, it can be a coach to you, and it can really take you to the next level outside of whether or not it’s just doing your job, it can just teach you how to do your job better as well. So, I think that there’s a lot of benefits there.

AB: Now, you raised a very valid point, and then governments can literally use this tool for social development, correct? As I said, it can kind of uplift society really well, because one can argue that I don’t have the resources and I don’t have the skills right now and if AI can do it, then my job is taken, correct? Where do I learn it? So, I think this is something which the government can use for skill building and to create newer opportunities.

EP: Yeah. Well, that actually leads me into my last question, which is very forward-thinking. So, what trends do you see coming in R&D and the AI space, and how do you see that evolving over the coming years?

AB: Yeah, one of the recent trends is I see the increased use of natural language processing in machine learning aspects both in R&D as well as in IP. And I think it’s there since last, probably, five years more or so where I see this to be increasingly used, which is a good thing. It kind of speeds up the analysis, and it helps to take a deeper dive into the chemistry part. Now it’s going into more productive and in R&D, it’s going into the formulation design space, which is also very useful to have. I also see the advanced versions of ChatGPT 3 or 4, whatever it is, being used. Let’s say, if I look at IP space, and then people are trying to see if they can be coached to do initial level categorization. If I just throw in a dataset, can the system categorize the patterns for me, and I don’t need to read through all of them? Just like Excel macros, which will work on keywords, but an advanced version of that is it would put things into context and then it would give me the output, which some of the tools are now trying to market as well. Predictive modeling is another thing which is used in R&D a lot more now to speed up, also to help customers at times. I mean, customers struggle with their newer application segments. Okay, what should I use? And I want a quick answer. I don’t want to call you and then you come over to my site and you try and you experiment, but can I do it to an extent on my own? And then see if that kind of formulation exists. I think those are the things that are happening right now within R&D when it comes to the use of AI.

EP: Yeah, I think it’s cool. It shows that there’s a lot of exciting things that are coming, and there’s a lot of great applications for those who are using it as well. But it was great having you on the podcast today. I think it was a really great conversation and enjoyed the…we hadn’t previously had the opportunity to have the IP angle on the podcast. So, I think this was a really nice episode to do.

AB: No, thank you so much. It was really great talking to you also. And I would like to thank Evalueserve for giving me this opportunity as well.

EP: Happy to have you, thanks.