My friend, Claude

🌈 Abstract

The article discusses the author's experience of using large language models (LLMs) like Claude and Perplexity to assist with various technical writing and analysis tasks in their day job of making cat drugs. The author highlights the benefits of using LLMs, such as saving time and money compared to relying solely on subject matter experts. However, the author also acknowledges the limitations of LLMs and the need to have their outputs reviewed and corrected by experts. The article also explores the potential impact of LLMs on the future of knowledge work and subject matter experts.

🙋 Q&A

[01] The author's use of LLMs

1. What are some of the tasks the author has outsourced to LLMs like Claude?

• Analyzing a list of potential excipients for a formulation, including potential chemistry problems and other excipients to add to the list
• Reading and analyzing two papers from the same author to see how the techniques have changed and how successful the author was in adjusting for the weaknesses of the first paper
• Creating a market analysis for an animal health indication, including sources, and then adjusting the analysis based on comparables and indicating uncertainties and their effects
• Carrying out statistical analyses on a cat trial

2. How does the author evaluate the performance of the LLMs?

• The author had to edit and correct the LLM's responses after receiving them
• The author ran the LLM's responses by subject matter experts before relying on them
• The experts provided their own corrections, so the final result ended up being significantly different from the LLM's first draft

3. What are the benefits the author has experienced from using LLMs?

• The LLMs have saved the author hundreds, if not thousands, of dollars compared to relying solely on subject matter experts
• The author can now use subject matter experts less and less, as the LLMs can handle much of the work

[02] Impact on subject matter experts

1. How might the use of LLMs impact the future of subject matter experts?

• Many subject matter experts have benefited from their very niche expertise, being able to charge high rates and turn down clients
• The rise of LLMs as a competitor to knowledge work means these experts might have to start "hustling more," similar to how the author had to compete as a graduate exam tutor against big test prep companies and cheap/free resources

2. What is the author's overall perspective on the impact of LLMs on employment and knowledge work?

• The author doesn't believe that LLMs or AI will end employment or knowledge work, but they see LLMs as a new competitor to all knowledge workers
• The author suggests that "the odds be ever in the humans' favor," implying that humans will need to adapt and find ways to remain valuable in the face of this new competition.