Working with AI instead of against it. This is what we have been doing since last year and we have founded the DR-AI-HM working group. This group investigates how we can safely and efficiently integrate AI tools into our work processes.
Among them is Ravdip Guron. As part of the working group, she deals with how AI tools could be used in Medical Writing.
Ravdip has been testing numerous AI tools in recent months and checking them on criteria such as safety, reliability, efficiency, etc.
Could the AI convince? 🤔
🔍 Literature searching tools performed best in Ravdip’s review. Platforms such as Elicit and SciSpace offer good support for literature research and usually get their suggestions from reliable databases such as PubMed.
✅ Text editing tools such as Paper AI or Grammarly are handy for quick text corrections and suggestions for improvement, and can also be installed as a Microsoft Word add-in.
🖼 Image generating tools made it harder to get good results. For example, in the representation of hands – sometimes a finger can be missing.
📃 The worst performing were text generating tools. Here the biggest shortcoming is the unreliable source. In addition, prompts must be formulated very precisely, otherwise the program will fill in gaps itself.
❗️ A major weakness that all AI tools have in common is insecure data processing. This represents a major hurdle, especially when processing sensitive information, such as confidential customer data. Compliance with intellectual property rights is also often disregarded, for example with tools where data is loaded in order to create a summary.
Ravdip’s conclusion 👉 Although AI tools offer some useful functions, the lack of data security and disregard for intellectual property rights still largely limits the possibilities of use in Medical Writing.
But we will definitely keep this topic in mind and are excited to see where the journey with AI takes us next. 🚀
Do you use AI tools in your company? Let us know!