A dream of science fiction writers until a few decades ago, artificial intelligence (AI) has become an integral part of our existence, with a particularly prominent role in healthcare – as a tool to both manage and analyse information and to assist diagnosis and follow-up in several therapy areas. Its involvement in medical writing was therefore bound to be the next natural step.
AI technology is especially useful for carrying out repetitive tasks with an elevated degree of redundancies, and for this reason it is already used in Pharma for the development of regulatory documentation, such as CSRs and submission packages. Most of the time and effort involved in producing these documents lies in collecting information from existing sources (e.g., study protocols, figures, tables, and statistical analyses) and packaging it under the correct section titles. This does not strictly require the scientific acumen and editorial skills of a medical writer; their input would remain essential when it comes to polishing the final product and providing high-level scientific interpretation.
By using a software process called natural language generation (NLG), AI can produce complex documents in a fraction of the time they normally require (days versus weeks) and therefore accelerate submissions and marketing authorisations, positively affecting the budget of both pharmaceutical and communication companies.
Like most modern technologies, AI is rapidly evolving, and is now being used to draft materials that require a higher level of editorial skills – such as peer-reviewed articles, abstracts, or posters – and its capabilities are increasing. Less than two years ago, a machine-learning algorithm called GPT-2, with a processing power of 1.5 billion parameters (configuration variables required to make a prediction in machine learning), was able to generate alternative paragraphs for an article published in the New Yorker. More recently, the updated GPT-3 version, albeit still in beta, has put its 175 billion parameters to the test by writing an alternative storyline to Kafka’s The Metamorphosis.
This begs the question: is AI bound to take over the job of writers (both creative and scientific), thereby dampening our imagination and making our editorial skills redundant? Or will it become a tool to save time on tedious tasks, help us overcome the infamous ‘blank page syndrome’, and allow a higher number of quality pieces to be published?
I’d like to think that the latter is a more likely scenario, and bigger minds have already started an exciting debate on the matter. So, where do you stand?