“What if a cure for an intractable cancer is hidden within the tedious reports on thousands of clinical studies? In 20 years’ time, AI will be able to read — and more importantly, understand — scientific text. These AI readers will be able to connect the dots between disparate studies to identify novel hypotheses and to suggest experiments which would otherwise be missed.
AI-based discovery engines will help find the answers to science’s thorniest problems.”
— Oren Etzioni
“Semantic Scholar” is a better idea than the REF. But wait, what do they have in common? Here’s what. Semantic scholar looks through existing scholarship and lets you use it and build up on it. It may uncover an article which no one has read many years after it was published, and let a new researcher learn something. It helps cut through the bullshit – perhaps not in the best possible way, but AIs are work in progress and will surely evolve into cleverer versions. The REF, conversely, fosters the production of bullshit. Its existence scares scientists into producing more crap (sorry, dear colleagues) because that’s the way in which we, and our university departments, are assessed. We all know it’s a game, and some refuse to play it, but people in their early careers have more incentive to play along than to protest by producing fewer, better pieces of work – even though it actually is in our long-term interests, we are fooled by fear. And so we write, and publish, instead of thinking and publishing less, better stuff.
If I were a science policy-maker, I’d put my money on tools that facilitate tedious, or downright impossible, tasks such as sifting, navigating and organising existing knowledge and debates. And I would leave scientists with a bit more freedom to actually think, be curious, produce ideas, hypotheses and sometimes even knowledge, and – very importantly – also to make mistakes in the process. Oh, and having access to jobs that last longer than a year or two before having to move continent with or without your significant other, or instead of deciding to have children, would help. But that’s another rant, perhaps for a future post entitled “If I were a science policy-maker”.
For more about Semantic Scholar, see: https://www.newscientist.com/article/dn28434-ai-tool-scours-all-the-science-on-the-web-to-find-new-knowledge/