Interview with Jackie Hunter, PhD – CEO of Benevolent Bio
Jackie Hunter gave one of the most interesting talks at the WebSummit this year. During her talk volunteers kept on putting boxes and boxes on stage. After about 5 minutes, and probably 100 boxes later, Jackie explained to the audience that these boxes represented all scientific papers published in one day. She cleverly showed that the amount of science data that is currently produced is just too much for any mere mortal to comprehend and use. Hence the need for AI, and hence her company Benevolent.ai.
I sat down with Jackie and asked her a few questions
Hi Jackie, thanks for meeting with me. Can you give me a little bit of background about yourself?
Thanks Patrick. I spent most of my career in the pharmaceutical industry. I was the head of Neurology and GI at GSK and I was responsible for all the basic research in that area, all the way through to Phase IIb clinical trials. Then I did a couple of years of consultancy, trying to spin assets out of pharmaceutical companies, but I found it very hard to align venture capital and pharma companies. Each time we got a VC interested, the pharma companies took the asset back in. Next, I ran the BBSRC research council in the UK, which was fascinating. This fitted well with my love of data because the BBSRC funded a number of big data projects, both in agriculture and bioscience research. This all came together when I got my job at Benevolent a year and half ago. My experiences in the industry had taught me how important it is to make good decisions. Yet frequently decisions in the industry are not as well informed as they could be. It is an evidence-based industry that doesn’t use as much of the evidence it could. Not that they willfully do so, it is just because there is now so much new evidence available that the traditional techniques to mine evidence are simply not good enough. Hence my attraction to Benevolent. They can utilize the new data and really think about how to be disruptive and change the drug discovery and development process.
Can you give me a one-minute pitch on what Benevolent is doing?
Scientific discovery, especially pharmaceutical industry, hasn’t really changed over decades. But the rate of generation of new data has grown exponentially with 90% of the new data being produced in the last 2 years. What we are doing at Benevolent is using AI to allow scientists to use all this information in a way that was previously impossible. We can mine hundred million documents and generate new hypotheses linking targets to disease and build new systems that allow us to produce better molecules, and more productive properties, in a 1/10 of the time it traditionally takes a pharmaceutical company to do. So overall, we improve the efficiency of the process and increase the success of the process.
What made you interested in AI?
I’ve always been interested how we can use information better. I’m naturally inquisitive and I remember when we used to have journals in libraries I was always looking at the odd, for example renal physiology journals, just to see if there’s anything that could resonate with what I was doing. And then, when I was running the BBSRC, I visited some centers like the supercomputer Center in San Diego, heard a lot about the real important data repositories in neuroscience and began to see how AI and machine learning was being used. So for me it was the ability to access all this information and to create new knowledge. The only way to do it is through AI machine learning. So it seemed like a natural progression. I had known Ken Mulvany the Benevolent founder before. I was a nonexecutive his previous company, so I had been in touch with Benevolent and had seen the evolution. The company actually started the end of 2013 but basically no drug discovery was done until the beginning of 2016. The first couple of years was really building the tech and building the specialized domain dictionary that allows us to recognize all the entities in the literature. To recognize a gene, protein, a compound, etc… and begin, through natural language processing, explore the published relationships and build a huge knowledge graph of these relationships.
How has the last year been for Benevolent?
It has been tremendous. Since I’ve joined we have in-licensed some molecules from Johnson & Johnson. The technology has come on in leaps and bounds. We published our patent for ALS where we have been able to demonstrate that we come up with novel hypotheses that have been validated in-vitro and in-vivo and we have got a lot more of patents in progress at the moment. The technology has really developed. A lot of that is through hiring drug discovery scientists, chemists and biologists. They are feeding that all the time into the tech. This rapid turnaround and feedback is vital because, as you know, technologist like to develop tech. Frequently technologists will come up with technology and look for a problem it can solve rather than laser focusing their technology on the problems that are really difficult to solve. And that’s what we do at Benevolent. The technology is laser focused on problems in drug discovery. Plus, they have to do that last 20% to make them reproducible, in production mode, and very user-friendly. Even I can use it, so it must be really user-friendly 🙂
Who do you follow for inspiration?
I follow number people on twitter doing exciting stuff like Atul Butte and Barney Pell. But I am also inspired by patients. I have lost friends to some serious diseases and I look at how people deal with chronic diseases like multiple sclerosis, Parkinson’s disease, etc. That drives me to come up with new solutions. We bring patients into Benevolent to talk about diseases. It is really motivational, especially for technologists, because they’re usually somewhat removed from what they’re trying to do, so it helps making it more real for them. We’ve actually changed our culture. We don’t talk about tech and bio, we talk that everybody is a drug discoverer.
What trends are you looking forward to in 2018?
It is going to be very interesting to see what’s going to happen with the applications in AI in general. Within the UK there is the AI Review and there is a commitment to create an environment which allows AI to flourish. Importantly, I think dialogue about transparency and AI pros and cons is important to have. One of the things that we can do with Benevolent’s system is that we can always can go back and see the trail that led to a hypothesis and the actual sentences and papers it came from, so we can make a judgment as to the quality of the hypothesis. So we try to be as transparent as possible. For the industry is going to be important to focus on problems, and generate solutions, that really matter for society. Whether it is policy or medicine. Because if people see the benefit of technology then they are much more likely to accept it. There’s a lot of hype of the Elon Musk’s of the world about the dangers of AI, and I don’t think we should ever underplay the risks, but on the other hand if we can show the benefits you can have that discussion about risk-benefit and much more informed way.
What can we expect from Benevolent next year?
We are going to start some more clinical trials and we will be able to talk more about some of the positive hypotheses that we generated. We will be ramping up our in-house chemical programs. So I think next year for us will be really exciting because it will be the accumulation of the past couple of years work and we can really start to ramp up our discovery engine. We will have some positive news flow coming out next year.
Thank you for your time and good luck with Benevolent: www.benevolent.ai