Consulting Services

Contact Me

The first session is free of charge. In this session, we will discuss your specific needs, if my services are truly required and if so the direction of our collaboration. Rates are flexible and negotiable for each situation.

You can reach me at fumiy@student.ubc.ca with the subject “Consulting Request”

Why might large-language models (LLM) not be enough for me?

To be honest, LLM tools (like ChatGPT, Claude, Llama etc...) are extremely useful tools. I use it day-to-day to help with my writing, coding, brainstorming etc. For most things, these tools (even the free version!) is enough. There's a reason the stock value of Chegg dropped significantly following OpenAI's first major release - it was like a better version of search engines like Google which actually 'understood' your questions and could provide answers as if you were talking to a humanized version of a textbook.

So... if LLM tools are so great, why might my services be useful? Let's dive into the theoretical limitations of LLMs - and how I can cover those gaps?

Theoretical Limitations of LLMs

LLMs are great text-predictors: it answers with text it believes is most appropriate for the situation based on the text which was used to train it. Another limitation of generative models - it 'responds' by generating text given an input - so there is a limitation in how much context and information it can 'remember' while giving responses. Therefore, it is a great query engine - but limitations exist in its 'problem-solving' capabilities. While AI agents are a trillion dollar problem which large tech companies are racing to solve, its current capabilities are still limited. So here I am!