Battling information overload: the AI-driven research assistant tool that can accelerate scientific breakthroughs
Do you ever get the feeling that there is so much information out there, you don’t know where to start? Or know how you could possibly ever take it all in? A diverse group based mostly out of Norway, Iris.ai is a team of experts, innovators, and programmers tackling this problem.
With more than 3,000 research papers being published every day, and 200 million already out there in the world, Iris.ai have created a system that aims to tackle this information overload. Their AI-driven research assistant tool scours papers for those relevant to you and your work, potentially cutting your research time down from 6 months to 1 month, meaning that breakthroughs can happen faster. By aiding scientists from virtually all fields in doing their jobs, Iris.ai believe the knock on effect of their technology in making the world a better place could be huge.
Here, they tell us more about their tool and how it works.
Where is your team based, how big is the team, and who is involved?
We are a team of weirdos and overachievers working fully remotely from across Europe. We are 21 people representing 16 nationalities, spread over 10 countries. The company was founded by: serial entrepreneur and Faculty in Artificial Intelligence at Singularity University, CEO Anita Schjøll Brede (Norway), as well as former investment banker turned serial non-profit founder President Jacobo Elosua (Spain), and AI Research academic turned startup founder CTO Victor Botev (Bulgaria). We met at Singularity University back in 2015 where we united around our common mission and set out to positively change the world through science.
Tell us more about the problem your AI seeks to fix.
We are living in a world with exponential growth of scientific knowledge, where human ability to keep up diminishes quickly. We believe the puzzle pieces of the answers to some of our most pressing problems are already out there – but no one human has the capacity to assemble it. We thus believe that by building a machine that can help us navigate, extract and make sense of all scientific information available. By allowing humans to have all relevant knowledge and its implications at their fingertips, we will enable those humans to solve some of the greatest problems facing humanity.
How does your groundbreaking technology work?
Our AI technology is mainly in natural language processing applied on scientific text (research papers, patents, clinical trials, and technical documentation). The Iris.ai technology uses key concept extraction, contextual synonym enrichment, neural topic modeling, and word importance-based document similarity, which allows us to build intuitively meaningful content-based indexes.
This poses a novel and more efficient way of performing literature searches and systematic research landscape mappings, reducing manual labor by 78%. The two initial Iris.ai tools for Academia, Explore and Focus, are already deployed at university libraries across northern Europe and are used regularly by around 20,000 students and researchers. In parallel, we are offering tools for Industrial R&D that help researchers pinpoint and extract precise key information – using technology that rapidly can be reinforced in the domain of the research.
What made you want to enter the $5M IBM Watson AI XPRIZE?
We started Iris.ai as a for-profit, for-impact entity. Bringing positive impact to the world has always been the driving force of what we do. That, and a fundamental love of science. As such, our goals aligned well with the focus of the $5M IBM Watson AI XPRIZE. The validation of being recognized as a participant and making it through each round validates both our positive impact and our technology contribution, and that means everything.
Why is what your team is doing important now and in the future?
Right now, we are enabling researchers to do radically better literature reviews and allowing them to spend their time doing other more important work. For example, our offer of free premium accounts to COVID-19 research has given our users the ability to do better research on topics from COVID-19 aftereffects on children, the impact of loneliness of isolation on mental health, job stress of frontline health care workers to the COVID-19 "infodemic" all the way to the effect of the pandemic on the stock exchanges and Bitcoin market.
Literature reviews are just the start. Our industry tools take it to the next level, with data extraction and identification. Ultimately, we envision an AI system that will be a valued team member in every research team, ensuring that the humans have all available relevant information at their fingertips at all times.
What team accomplishment are you most proud of?
In early 2019 the company hit a couple of road bumps and we found ourselves on the verge of bankruptcy. With two weeks of salary left in the bank we had to ask our 15 employees to make not only a sacrifice of reduced and unstable salaries, but also a leap of faith, and give us four months to save the company. We navigated those months with the entire team's support, and came out on the other end with a research grant, a major client and investment – and a team where we knew everyone had our backs. As founders, we can't even begin to describe how proud we are of having built a team like that.
How has the IBM Watson AI XPRIZE competition furthered your success on top of that?
XPRIZE has given us a solid amount of credibility - both on the impact and the technology side. This has strengthened our applications for grants and conversations with investors, and with every new milestone we got accepted through, our confidence in what we were doing would rise. This changed us as a company, and each of us individually as co-founders.
Outside of your work, what's an area of AI that's exciting you right now?
I am really excited about the much more nuanced conversations that are taking place around the major tech giants and their role in spreading information and misinformation. The interface between AI, corporate policy, law, ethics, and truth is hard, complex, and fundamentally important to humanity. These conversations were not possible outside of a few small special-interest groups just a couple of years ago, but now people are starting to catch on to what these self-reinforcing filter bubbles are and what role machine learning plays in exacerbating the issues. I absolutely love technology, but I love it even more when we can have the necessary and hard discussions around it.