Events

A workshop on the use of Artificial Intelligence (AI) for scientific writing was held between 21st and 22nd November 2023 at the University of Malta Common Room, as well as online. This occasion marked the first event of the ERA SHUTTLE project.

The workshop titled, “Scientists Need More! Next Level Productivity Using AI Assistants” was the first event of the project, and was hosted by Daniel Mertens, on behalf of the project partner Schiller & Mertens GbR.

Through an interactive approach, he engaged over 50 attendants who were both in-person as well as online attending through Zoom on the use of several AI tools to boost both the productivity by around 25% and quality of output by around 40%.

Following an introduction which demystified how generative AI tools such as ChatGPT operate, the attendants were made aware of the different strengths and weaknesses of different AI tools. Certain tools such as ChatGPT were noted to be better at generating content, however was not as adept at finding information, in that case tools use as Perplexity were more appropriate.

When it came to seeking out information required by scientific writers, then another set of tools (Elicit, Litmaps, Research Rabbit etc) were found to be more appropriate.

In addition to exploring the different tools, attendants were also taught how to prompt the AI in order to have it respond in the most efficient, direct and clear-as-possible manner. This made it possible to engage the AI in decision-making, by having it use methods such as ‘Second Order Effects’, or ‘Regret Minimisation Framework’ among others. 

The workshop also explored different use cases for AI tools, such as its utility as a brainstorming companion, however, a number of risks were also discussed. 

To date there continues to be the risk that certain AI tools have a tendency to ‘hallucinate’ and provide convincing information which could in fact be false. There also is also the risk that new and original information which is fed into the AI by a user can be regurgitated by the AI itself to another user who could be a competitor.  

Therefore, attendants were taught to be prudent and cautious with how they make use of the tools to avoid using false information and to avoid losing confidential and proprietary information.

Workshop attendants were also given a crash course on how to use a mix of ChatGPT and Perplexity to create a programme in Python, and also use the tools to troubleshoot and solve errors that arise in the process, without having any prior experience in programming at all.

By the end of the workshop, effectively all of the attendants remarked that they will in some manner incorporate one or more of the AI tools which they learned during the two days. Many of the attendants even said they will use the knowledge to teach their peers and colleagues on how to get started, in order to facilitate both of their work.