Feedback from an internship on LLMs

Published at 2023-07-09 by Ana Rodrigues

On our last blog post we shared with you that we were welcoming UvA interns at Impacter to help us look into Large Language Models, and if and how we could use them to improve our tool.

Today, we are looking into their feedback about the experience and how challenges can be addressed at a company setting. According to Julia Pestel “Although ChatGPT and other LLMs are everywhere at the moment, and used by a myriad of different parties, I never realised how little I knew about them before starting this assignment. We did research into LLMs in general and how we would envision the use of these models in a software like Impacter. Moreover, we learnt about prompt engineering and experimented using different prompts in LLMs. We implemented a very different way of thinking and working than what we’re used to at the UvA which was challenging, but very rewarding.”. Marijn de Gans mentioned that “For four weeks we focussed on the possible challenges that upcoming Large Language Models would pose for the company and how we could best anticipate them; can we use LLMs, like ChatGPT, to our advantage, and how? (…) The hands-on experience greatly enhanced my understanding of how complex challenges can be approached and overcome through a systematic and empathetic approach.”.

If you also are curious about Impacter, have some creative ideas that could be implemented in our tool or simply would like to have an experience in a company setting, don’t hesitate to reach out.