Today AI technologies are employed to both create and consume research. Scientists across the world are using AI to be able to conduct research that will solve the world’s most urgent problems—a goal they are able to march toward only by assimilating content that is sifted through for relevance and routed their way by powerful discovery algorithms. In the scholarly kitchen, AI is at once the main ingredient and the serving platter, and researchers, both the chef and the diner. It is this all pervasiveness that pre-defines any exploration of the role of AI in research.
The single biggest promise of AI is its potential to democratize research by lowering the barrier to access must-have support. Studies show that the average researcher is likely to spend four hours searching and up to five hours reading per week. Imagine an algorithm that curates a content feed matching the researcher’s interests, runs through each paper in it, and suggests summaries for each that she could quickly digest before deciding whether to read through 4000 words of the full paper. Now consider the fact that such personalized support is at her disposal at no cost.
Like a handful of others in a mix of tech startups, nonprofits, and behemoths on the AI solutions provider landscape, this is just one way in which CACTUS is challenging the status quo across the research workflow and helping redefine how science is disseminated and digested. A growing digital maturity buoyed by large and diverse training data sets, quantum computing power, and advances in natural language processing is allowing us to design an ecosystem of interoperable tools that support researchers every step of their way. Here are some ways in which such support is helping transform how research is delivered to the world in various ways:
Synthesizing insights from big data: The ability to extract insights from big data linked to real-world evidence is a key driver to decision-making in pharma research. Machine learning allows synthesis of actionable insights from continuously expanding content corpora, which in turn informs R&D spending decisions to the tune of hundreds of millions of dollars.
Driving trust in science: To build trust in science, AI technologies are being used to assess the validity of scientific reporting and provide a quality standard that serves as an unbiased supplement to the human-based authentication that peer review offers.
Improving current journal publishing workflows: An engine trained on millions of published articles such that it can instantly identify the most common errors and omissions in a manuscript before it is submitted to a journal can ease up and hasten manuscript preparation and reduce the chances of rejection.
Enabling new business models for scholarly publishing: Researchers spend a lot of time identifying a journal to submit to because they can only submit to one at a time. If rejected, they have to start the process all over again. An AI-powered article marketplace flips this limitation on its head and allows journals and researchers to choose each other on an open platform.
In research as in any industry, the essence of disruption is defined by the criticality of the problem(s) solved. As the technologies under the AI umbrella expand in scope and depth, their impact will become more meaningful and relevant to research dissemination and consumption. It is reasonable to expect that the coming years will see AI investments in research being increasingly focused on the consumer side. There will emerge challenges in training (quality of data sets) and deployment (data ownership and privacy), but there is little disputing that technological advancements will be of particular consequence to researchers from the Global South where R&D budgets are tight and research output is witnessing strong growth.
Put simply, it is AI that promises to level the playing field in a manner that researchers of tomorrow are defined by the strength of their ideas, and nothing else.
A slightly modified version of this article was originally published in Express Computer