Should we be using ChatGPT?

Published at 2023-06-20 by Ana Rodrigues

Conversational Large Language Models such as ChatGPT show the revolutionary progress in the field of AI recently, and cause entire industries to stop and think about their added value.

At Impacter, we’ve been building software for years that provides automated feedback on research proposals. Researchers use the tool to avoid common mistakes, and improve their chances of receiving funding for their research project. Up until now, the role of the software has been to spot the problems, and leave it up to the researcher to find a solution. So for instance, a researcher could write: “We will actively involve our key stakeholders in this project”, and Impacter would highlight ‘Actively Involve’ to ask: What does that mean? How will this involvement take shape? Have you safeguarded this in your budget? Similarly, for Key Stakeholders it will ask who those are, and whether the researcher could name them.

As one of the next steps for Impacter, we are considering to use the existing tool in combination with Large Language Models such as GPT to be able to suggest some solutions to challenging parts of the proposal. Implementing this has moral, legal, technical and practical implications and there could be some important limitations.

In order to learn more about this and help us making the decision of testing and using ChatGTP, we are welcoming, during June, three interns from Cognition, Language and Communication studies in UvA who will be working on this as part of their assignment. Visit the blog again in a few weeks to read more about their experience and conclusions.