Scholarly publishing has changed considerably in recent years. Perhaps the biggest transformation is one we now take for granted – the move to digital. Publishing and scientific R&D are not the only industries to have witnessed the remarkable impact of digitization on how people work. But for our industry, this development is especially important due to the sheer volume of data researchers must sift through. The growth in the value of indexing, for example, alongside various forms of analytics, has dramatically changed scholarly publishing for the better. In a world where approximately 2.5 million new scientific papers are published each year, digitization has become the imperative.

Digitization, when supported by the right analytical tools, has made it far easier for researchers to find answers and insights into increasingly complex problems by filtering out the unnecessary data from the huge volumes available to them. To find out exactly what the right analytical tools are, many companies have invested in Artificial Intelligence (AI); with analysts Forrester estimating that the AI market grew by 300% in 2017. Technology, and particularly AI, is leading to the emergence of the ‘augmented researcher’.

From human curation to machine reasoning

As scholarly publishing has become a digital enterprise, the move to create semantic data that captures knowledge has increased significantly. Another way to describe this trend is as a shift in focus from reading articles as a whole, towards finding individual, semantic ‘facts’ reported in publications. This has been driven by the maturing of automated approaches to identifying and extracting these facts; as well as the steps to bring AI into fruition in the form of machine learning. In essence, recent years have seen a step-change from human curation in isolation to rules-based automated indexing approaches,  and then to the applications of statistical approaches such as deep learning and machine reasoning. These approaches are helping researchers to access insights in a far shorter time period, greatly improving productivity.

Semantic data is important to R&D, because it means we can link facts that are related across papers, and over different domains of knowledge – enabling us to deliver insights that might not be obvious from reading one paper alone. To do so requires normalizing the terminology with taxonomies, to allow a network to be created. The increasing reliance on linked facts mean the demands of the modern researcher are changing; researchers need bespoke analytics products for their specific needs, built on robust semantic databases. In today’s world, this more often than not means solutions that combine semantic technology methods, augmented with machine learning and machine reasoning approaches.

As our understanding of AI grows, how we apply it will become even more sophisticated; and thus, even more beneficial to researchers. This could mean extracting facts from various scientific literature sources to highlight and deepen existing knowledge, or focusing on how AI can be made more efficient with approaches such as transfer learning to identify new knowledge. At the same time, AI tools will also become easier to understand. At Elsevier, for example, we are currently exploring something called ‘AI neuroscience’ – that is, building tools that look inside the ‘black box’ of deep learning models, and working out how an individual decision is made.

What to look out for over the next 12 months

In 2018, thanks to such developments, we will see a greater interest in and acceptance of AI as a valid technique in R&D. At the same time, as technology makes the world ‘smaller’ and continues to improve how we communicate, 2018 will also see a greater propensity amongst scientists and researchers towards working together and forming partnerships. The insights we gain will then be shared through collaborative networks. As a result, of the progress in semantic data researchers will increasingly benefit from broad, integrated, cross-domain networks of linked facts. This will allow them to draw inferences and identify patterns from the use of machine and deep learning – which are fundamental aspects of artificially intelligent systems.

These techniques offer the ability to aim AIs at problems we are interested in solving, and having the means to understand and interpret the answers the AIs provide. Together, these two factors of growing interest in AI and greater collaboration will become increasingly important if we are to overcome the productivity crises that many disciplines of science and research are experiencing. Research at Stanford University has indicated that since the 1930s, the effective number of researchers at work has increased by a factor of 23, but annual growth in productivity has declined. As a result, new ideas are becoming more expensive to find; using AI to augment researchers will be a key weapon in the fight to overcome these issues.

The building blocks for an AI-driven future are already in place – with the general transformation to digital underway and a willingness to work alongside, rather than in opposition to, machines. I believe, in the next 12 months, we will begin to see the first fruits of these advances in the use of AI to augment research; helping us to make meaningful progress towards reversing the productivity crisis in science and R&D, and improving outcomes for humanity by solving the problems we face globally in many areas – from antibiotic resistance, to environmental degradation and climate change.