Sabine Loss

Large language models are not new, they are around for about 4 years already, but last year, ChatGPT started a hype around large language models. The hype is closely connected to the specific capabilities of ChatGPT (or GPT 3.5, the underlying large language model) in the area of text understanding and generation. This also puts a spotlight on technical writing and how large language models can change the way help content is created and presented to users. In this presentation, we want to look under the lid of large language models for a better assessment of what they can do, what they can't, and also the costs involved. We will then look at some of the use cases that are currently discussed and whether they make sense from a user perspective, but also from a cost-benefit analysis. We'll end the presentation with some of the use cases we think can make sense to evaluate further.

Highlights: There is a quite a hype around the topic of ChatGTP these days. Everyone seems to talk about how it can revolutionize the way we create and provide help to the users of our software. But ChatGPT and Large Language Models in general are not magical tools, but it's all about mathematics and training.
In this session, we will
• Open the LLM's lid and have a closer look at how they work
• Discuss the hype, the opportunities, but also the myths of what LLMs can do for enabling users
• Share some use cases where LLMs can support technical writers