How AI is reshaping technology transfer, with RocketSmart

In this episode, we explore the world of tech transfer and innovation with Dvorah Graeser, Founder and CEO of RocketSmart. We discuss the commercialisation challenges universities face, explore the potential of AI in supporting technology transfer and highlight the legal and ethical concerns around using such tools in commercialisation more generally.

Also available on:

Transcript

Balzhan Orazbayeva:

It is my greatest pleasure to welcome Dvorah Graeser, Founder and CEO of RocketSmart. We will learn more about the company, and Dvorah will be sharing with us her extensive knowledge on AI and how it all relates to the high education ecosystems, including intellectual property, innovation, and all of the nitty gritty behind it.

I just would like to remind everyone that this event is part of our free Fireside Chat Series and for other events, including workshops, how-to sessions, head to www.uiin.org/events to see what is on.

Dvorah, thank you so much for joining me. It would be good if you could tell us a little bit more about yourself and then we can delve right into it.

Dvorah Graeser:

Absolutely. Thanks so much for inviting me, I always love talking about AI and tech transfer, two of my favorite subjects. I’m Devorah Graeser, Founder and CEO of RocketSmart, “Rocketing your IP out of the university into a great licensing deal”. I have been programming since I was 16, PhD in pharmacology, a program for the Human Genome Project, and then I became a U.S. Patent Agent because I was passionate about helping innovators protect their innovations.

From there, I learned that innovators also needed help with understanding the best business model. You need that for patents, but quite frankly, you need that for your ideas to make a difference in the world. For them to be commercialised, they have to fit into the marketplace, so I created our first AI models, basically to replace me for helping folks understand what the best business models are.

AI is a hugely impactful tool. It can make us much more productive, but one of the things I have learned is AI makes everything go faster, the good and the bad.

In order for AI to impact us in a way that we want, we have to understand the problem that we are trying to solve and how we want to solve it, and then integrate AI into our work process.

Balzhan Orazbayeva:

Very curious to learn more about it. Maybe just to start with and just to set the stage, how do you generally see AI, and other technological developments, transforming the traditional model of tech transfer in universities? Do we see many universities already picking up that trend? And maybe how do you usually help them with that?

Dvorah Graeser:

I have seen quite a few universities and other research institutions starting to do experiments with AI, in particular generative AI, so when I talk about AI today, I’m mostly talking about generative AI. These could include general tools like ChatGPT, Claude, or the specialty tools for tech transfer, like First Igniter Scale. IP generative AI in particular is a game changer because it is so very powerful.

The experiments that folks are doing in universities and other tech transfer organisations is to, for example, find industries where their innovations could be useful, but also to find good contacts within particular companies. Many times, an innovator will have this great idea, a university will have perhaps a great new drug, a new medical device, a new idea for a car, etc. Any of these ideas on its own is not really useful unless you can find the right company in the right industry at the right time. And then not only that, find the right contact and give them the right information in the right way. Talk about trying to find a needle in a haystack.

It is super hard for a human being to do. AI can be really helpful because AI can do a lot of the work and bring information to us, but because AI can bring us so much information, part of what I see universities and research organisations struggling with a bit is curation.

They get so much output from AI and the question is then what to do with it. Having a really good process can be helpful, and that is one area that I do see Tech Transfer Offices really working on right now.

Balzhan Orazbayeva:

That is a very good point because what we have seen is that often AI or generative AI as well can sometimes be seen like a black box itself. Being confronted by so many different outputs is one thing, but then also dealing with the tools such as AI can also be quite overwhelming, I can imagine, for some of the folks that work at the universities, especially if they are not experts.

So maybe the question is how important is transparency in AI tools for tech transfer? And for example, how do you help them curate that process and provide that transparency in the first place?

Dvorah Graeser:

I believe transparency is really important for any technological tool that we are using. We should be able to understand how it works, what data was trained on, how the data was collected, and then how it operates on information that we bring in in order to create its outputs. Unfortunately, the majority of generative AI models are not really explainable.

Explainability simply means: Do we understand why the AI does what it does? And the short answer is no, we don’t understand that. There have been many papers published on that, and I have yet to see a really good explanation.

But transparency also relates to how the data is collected, how our data is collected. If we are putting an idea into an AI tool, we should know what happens to that data, and that isn’t always clear. The policies of these companies, in my opinion, could be much clearer in order to provide that kind of transparency.

So yes, we need transparency but unfortunately, no, we aren’t always getting it. So then we need to ask ourselves is what is the policy of our organisation? For example, can we put in confidential data into one of these tools? But then, what is confidential data? You might say “Patient data is confidential”. What about personal data? What about other information? What about an invention that we have filed the patent application on, or maybe one that we haven’t, and that hasn’t been published yet? All of these different gradations of confidentiality and transparency are really important.

If we, as a group, demand such transparency and set guidelines for what we expect from our AI tools, we will get more answers. Sadly, at the moment it is really hard to actually tell what is going on with the tools and the transparency could be better.

Now, I do guide my clients in looking at the tools. I will contact the companies, I will read through all of the very long, fine print to understand what is going on. And I will also look at what others have said in the field, because there are other experts who have looked at these tools and how they are working to try to get information. But no one right now has a complete picture on transparency with AI or even explainability.

Balzhan Orazbayeva:

What would be the process for you to work with the universities with all of this very complex picture? Again, I’m also not an expert on AI, but I’m very curious to hear step-by-step how usually the work would look like for you when, say, you work with University A that has a particular, objective B when it comes to commercialisation.

Dvorah Graeser:

The first thing I do is ask them to describe their process to me and where the pain points and the friction points are. Where are there blocks in the process? Where do they feel they are spending a lot of time and quite frankly not getting a lot of return? Those are the areas that are particularly ripe for applying AI.

Now, I would ask that of them even before we start talking about AI tools and the reason why is because there are a ton of AI tools out there, but if I don’t know the problem that they really need and want to solve, then I can’t recommend any AI tool. It is not just our AI tools that I will recommend, I will also recommend general-purpose tools, specialty-tech-transfer tools, or maybe specialty tools that aren’t just for tech transfer, but that could help them overcome blocks in their flow and to enable them to move more quickly and more efficiently through the process.

I also need to hear about their frustrations because quite frankly, there are some things that only take time, but drive everyone bonkers. Those I think are also the aspects that are really ripe for AI. Now, it is even possible to bring AI on premises, that is, within a private cloud controlled by the university or other organisation, and to have the models operate in a particular way.

I know it sounds far-fetched, but it is possible. All of this comes down to one single question: What is the problem that you, as the university, as a tech transfer organisation, are trying to solve with AI?

Balzhan Orazbayeva:

Do you think that the universities versus, say, other research organisations or companies as well would have the same challenges when it comes to tech transfer and how would you usually go about their issues?

Dvorah Graeser:

Every organisation is different. You can’t even say that all universities are the same because it is not correct. First, different universities have different fields of technology where they are getting the innovations from their professors. They have different levels of staffing.

Some universities have people who specialise in small molecule pharmaceuticals. Other universities may have one person who handles everything that is medical, and maybe also robotics and, I don’t know, aquaculture. Depending on the staffing of the university and the variety of technologies that they are looking at, they are going to have different needs.

That, of course is different from a hospital. Hospitals tend to have very medically focused innovations, obviously. They are talking about a narrower range but, potentially, more depth to each of the innovations and also potentially a longer period for connecting them to a company because these innovations can be riskier.

A small molecule pharmaceutical is very risky. It can take longer in tech transfer to connect them to the right company who wants to take that risk on and they may have to do more work first.

Now, looking at companies – larger companies, smaller companies –, they also have wildly different needs. It depends on the area they are in. It depends on, are they competing on innovation? Because if they are competing on innovation and not cost, then I would say, “Yes, you do need to come and talk to universities, hospitals, and research organisations through an organisation like UIIN in order to be able to compete effectively” because these days competition is moving fast and it is global.

Balzhan Orazbayeva:

I couldn’t agree more. Especially for the smaller players or the hidden champions, it is always a tricky part to then be able to get to the forefront of that. With that, I also wanted to ask about how would you go about, not only identifying the IP, but also protecting it?

Because I think this is also one of the biggest challenges in tech transfer more generally. Is there any specific function or a specialisation that you would provide when working with clients all across the globe?

Dvorah Graeser:

Absolutely. When it comes to AI, there are a few things we would look at. Sometimes AI is a good tool, but sometimes also AI can complicate the process of the invention itself. For example, let’s suppose your researchers used AI to create their innovation. The U.S. has a particular set of rules and other countries or regions, such as the European Patent Office, are developing their rules for:

When is an invention not sufficiently inventive because there wasn’t enough human conceptual work or enough human ingenuity that was being brought into it? That is one place where AI can actually cause friction.

Of course, protecting AI itself is quite difficult because defining an AI invention is also really hard.

On the other hand, AI can also be really beneficial. For example, I would use AI tools to understand what the patent landscape is. Who are the competitors? What kinds of industries could best benefit from this innovation so we can tune and focus the patent towards those different areas? AI in that case then is really a big help for us, but we do need to be cognisant of all the ways that AI can influence an innovation, all the ways it can help, but also all the ways that it can hinder and make certain with our IP strategy that we take all these things into account in order to make the strongest patent portfolio for a university or other research institution.

Balzhan Orazbayeva:

I’m also very curious to hear, Dvorah, if you could maybe tell us a little bit more of the evolution of your previous organisation, I believe it is called KISSPlatform. How did it evolve into what you are doing today at RocketSmart?

Dvorah Graeser:

We actually started in January 2020. Guess what was just about to happen then? Yes, I was in Vietnam hearing about this virus that seemed to be spreading, and no one really knew anything about it. We originally had started to help start-ups define their business model. Because, again, I had learned that all kinds of innovators, from small startups to bigger universities to really huge corporations’ innovators, had a hard time defining their business model.

So, we started off with KISS Platform to try to help, particularly start-ups, define their business model so they could create a product that would fit into a particular market. However, what we found through the pandemic and afterward is that we are getting approached more by universities.

Now, I had been working with universities on patent protection. For many years. I’ve been a U.S. Patent Agent since 1996, it has been a long time, and I help them also understand a bit about the business and the market aspects in order to make their patents stronger, but we are being approached by them more than the start-ups to help them understand.

You can understand why universities have a wide variety of innovations. A startup maybe has one product, and they can do the research and understand how that fits into the market if they want to be successful, but universities every single day can get a new innovation that can fit a new industry and a new market with a new profile to create a new product. That is very difficult for their tech transfer offices to handle effectively, so that was why they started coming to us and how we actually then became RocketSmart, “Rocketing your IP out of the university and into a great licensing deal”.

One of the aspects that we focus on right now is partnering. We help universities, particularly those with older patents that have been on the shelf for a while. They are maybe seven years old in life science, but they are really great, and they are just sitting there and nothing’s happening with them. We helped find an industry partner, a company who wants to take a license to these patents for a variety of reasons. Maybe they don’t want their competitor to get a hold of it. Maybe they want it because, oh wow, they just realised they have developed an innovation, your university patent covers it and they want to be able to protect it.

That is what we have specialised in, but we do also work with university spun-out start-ups to help them find another company to partner with for commercialisation. We look at partnerships and relationships, this is not purely transactional. This is something which will help the university and the company together change the world, so we want to help them develop a longer term relationship.

Balzhan Orazbayeva:

I’m very glad to hear that you are also taking this sort of partnership, relationships-building approach, because I think one thing that is often being being said or one of the ways to criticise commercialisation is that it’s very one directional, very transactional, so it’s good to hear that you are also looking into how data-driven approaches can help in finding the right partner.

Do you find any challenges in this partner bit in particular? How easy is it usually to find using smart approaches when it comes to identifying partners?

Dvorah Graeser:

It really depends on if there are others in the industry commercialising technology as it is described in the patent or a similar technology. What can happen with patents is they can, let’s say they become more than 10 years old. At this point, if no one is commercialising technology, then the university or tech transfer organisation really has to ask themselves, “Do we want to continue?” Do we believe in this enough to believe that, before the patent expires, there will be someone who will want to license it and who will want to work with this particular innovation?”. That is a question for the tech transfer organisation.

When no one is commercialising at least similar technology, it becomes really hard to find a partner because no one is working with this technology to make products

If no one is working with a technology in a business sense, even if it is doing business-oriented research, it is really hard to make that connection between the patent holder, the tech transfer organisation and the company, because then you are selling them on, not only the innovation, but the idea of coming into this particular area, and that is super hard.

Balzhan Orazbayeva:

What are some of the key legal and ethical concerns? I can imagine some that universities may have to address when applying AI-driven tools to tech transfer in particular, but also partnering more generally, when it comes especially to IP ownership, because this is something that we have already discussed.

So, just curious whether you have any advice when it comes to taking some very concrete steps towards that direction.

Dvorah Graeser:

There is always going to be a certain element of risk when using an AI tool because you are putting your invention description out there into a software platform. You want it to be handled confidentially and privately and with respect, but you don’t actually know even if the company has a good policy.

It’s still taking a risk, so one thing that I always discuss with my clients is “what kind of risks are you willing to take now?”. For example, many of my client universities, but also small businesses, have said: “Once we have filed the patent application, that is fine because we know we have the date, and we know we are protected”.

Others have said: “Once the paper or the patent application is published, that is great”. Others are also willing to work with these tools, even at an earlier stage, because they say: “we are a small office, we don’t have a lot of resources. If we don’t take some more risks with our IP, then we may not be able to find partnership opportunities for the patents and in the IP will never change the world or make the world a better place, and that is not something that we want. Our mission is to get our inventions out there”

It depends on how much risk you want to take and, of course, once you filed, or once it is published, people can feel a lot more comfortable. In some cases, universities are even saying they want to keep some things as a trade secret. For example, a chemical process, they may not want to publish that, and they may not want to put that into a patent application, then that could be much riskier to put into one of these generative AI tools because the truth is you never really know.

That is also why I think it is important to discuss it as an organisation, as a tech transfer office, discuss it with your university or other members of your organisation. Make certain that everyone knows what can go into a particular generative AI tool and what can’t, and how to distinguish between different generative AI tools.

One of the things that worried me the most, I was having a great discussion with the university and they said: “We don’t put any confidential information into ChatGPT. So, I said: “That’s great. What is confidential information according to you?”, and they responded: “We know it when we see it”. And I said, “Oh dear. This is going to end in tears, I’m quite certain”.

Having policies, having definitions and making certain everyone agrees on what those policies and definitions are is the best way to go forward.

Balzhan Orazbayeva:

I wonder how it is in your work, because what we often see is that AI is often being treated as this unknown something and some people really embrace it. Some organisations really embrace it, make use of it, and use it for ethical reasons, but then some are quite scared of it.

Now that you work so much with the universities, I’m generally curious to understand which universities are more risk-averse and which are ready to take that risk and actually make use of such novel tools.

Dvorah Graeser:

Actually, it’s interesting that you say that because I would have assumed that, for example, smaller organisations, small universities would have fewer problems, and that kind of is the case. But I have run into some small universities where there is a blanket rule: “No AI, no generative AI of any way in the office”.

I’m like, “okay”, but then sometimes at some of these organisations, the employees will still use it, they just won’t tell anyone.

Asana did this great research project, and they found that up to 80% of employees use generative AI tools without telling their boss which, in my opinion, is a recipe for a disaster.

It is more about how risk averse the particular university is, how much they have discussed it with their legal counsel. Because there is liability that is potentially there. I also believe, and I think I understand why hospitals and universities, like a strong medical school, are more careful. A lot of the data that they have access to is super sensitive. They do need to put up more guardrails. Otherwise, I would say it depends on risk.

There are some organisations that say “no generative AI”, some that say “use it however you like”. I haven’t seen as many yet with really defined policies, but those are starting to come online.

That is really where the sweet spot is, having a policy, defining the policy, and then also letting folks know which generative AI tools are okay and which are not.

Balzhan Orazbayeva:

Maybe on that, I will steer our conversation towards even more forward-looking topics and looking ahead. How would you envision the role of AI or any other new technologies and tools evolving in this broader innovation
ecosystem?

Dvorah Graeser:

Generative AI will do two things for universities and organisations, particularly for tech transfer organisations:

1. It will reduce friction. It will help universities build relationships with companies with specific contacts in those companies to bring them the right innovation at the right time, packaged in the right way so that their contact in the company can really understand that innovation and can quickly decide yay or nay on it. It is going to speed up timing and efficiency by reducing friction.
2. It is also going to increase expectations. All technologies do this, if you think about it. Once we had the lightbulb, we expected our homes to be brightly lit all the time. Once we had spellchecker, we expected our documents to not have spelling mistakes, or at least the people would check before they are being sent out. So, generative AI in particular is going to raise expectations.

Universities and other tech transfer organisations that don’t use generative AI will get left behind because they won’t be able to fulfil the expectations of their commercialisation partners.

The same is true of small and large companies who want to bring innovation into the company, they are not going to be able to fulfil the expectations of the universities using AI if they themselves are also not using AI to bridge those gaps.

So, reducing friction, great, but increasing expectations. We have to watch out for that. If we don’t use AI, we could get left behind.

Balzhan Orazbayeva:

Very nicely said. Maybe we can finish off with your advice to those universities and research organisations that are perhaps hesitant right now to adopt any of the AI-driven tools like RocketSmart, for example, or any other ones for their tech transfer efforts. What would you tell them?

Dvorah Graeser:

I would recommend, of course, reading more about AI, talking to other members of the organisations like UIIN in order to learn from each other.

We shouldn’t all be reinventing the wheel. We should talk with each other and share our knowledge.

Beyond that, I think it’s good to pick a small pilot project. If you look at your overall tech transfer process, what is the one thing that has a lot of friction that slows you down that makes you and your fellow colleagues crazy? Pick that one aspect and then figure out how generative AI can help. Now I’m hoping someone will put my email address in the chat. I’m always happy to have a quick call with folks and point you to resources.

There are a lot of great resources out there to help you, but the single best thing to do is to understand the problem you are trying to solve and make that problem as small and as specific as possible for your first pilot project. That way, you are able to ask how AI can help you instead of fearing AI as a giant kind of blob, as you mentioned, the black box. Let’s transform that back into a tool that helps us.

Balzhan Orazbayeva:

Wonderfully said, Dvorah, and thank you so much for this conversation. I would certainly invite everyone to get in touch because I’m sure there are many more questions that could be explored in that area.

Thank you so much for joining me. I’m looking forward to continuing this conversation at a different stage as well. Thanks everyone for joining.

Dvorah Graeser:

Thanks for having me.

Balzhan Orazbayeva:

Thank you.

Ready for more?

If you are interested in how AI is revolutionising other aspects of higher education, head now to our episode Learning with AI: Balancing technology, pedagogy and ethics, where we invited Carla Aerts, Strategic Advisor at Erasmus X Innovation at Erasmus University of Rotterdam, to explore the potential of AI to reshape higher education, balancing the integration of advanced technologies with ethical considerations.

Want to know how other universities are facing their technology transfer challenges? UTS Rapido is an innovation hub from the University of Technology Sydney that is redefining how universities support industry with R&D innovation strategies. Learn more about it in our episode Commercialising innovation: UTS Rapido’s impact on research transformations.

Stay tuned for the next episode on this series and don’t forget to follow us on your preferred podcast platform!