In this final episode of Season 2 of The Puzzle, Isaak and Bruno tackle a question that’s more timely than timeless in our era of hype surrounding the topic of Artificial Intelligence (AI) and Large Language Models (LLMs): how do LLMs impact development projects?

Having used LLMs since the day ChatGPT became widely available in late 2022, I share my experience utilizing LLMs for product development, product management and product marketing use cases–as well as the benefits of using LLMs (both self-hosted and proprietary/cloud-based) to kickstart your research and learning.

We deep-dived into the impact of large language models (LLMs) on development projects. We explored how LLMs, such as ChatGPT, can be valuable tools when used wisely. I shared my experience of using ChatGPT for tasks like generating marketing copy and structuring unstructured data, while Bruno expressed cautiousness due to a preference for knowing the origins of information. However, we both acknowledged the utility of LLMs in aiding learning and exploration, particularly in development projects. Both of us emphasized the importance of understanding the tool’s limitations and critically evaluating its outputs. Overall, we concluded that LLMs can indeed be powerful aids for those who remain curious and willing to learn.

The duo highlights the importance of understanding what you’re looking for and having sufficient insights to evaluate the outputs. They emphasize that LLMs are not a source of knowledge but rather a tool that requires proper prompting and thinking through questions to increase data quality. Bruno emphasizes that one needs to be aware of vaguely at least of the expected outputs, content type, or else risk getting wildly incorrect tensors. The conversation highlights that LLMs can be an excellent tool for those who are naturally curious and intellectually curious, serving as an antidote to resistance to learn more things through exploration.

Also see the blog post with the distilled practical use cases of LLMs in development projects .