Transcript#
This transcript was generated automatically and may contain errors.
Two AI products. They're built on top of the same base model, so they must be the same, right? Well, not exactly. AI products may be built on the same base model, but the product developers decide what instructions they get and what tools they have access to, and tools are where the magic happens.
Because at its core, an LLM alone is just text in and text out. It can write you Python code, but by itself it doesn't have any way to run that Python code. It's just text. A tool is sort of like a predefined function that the model is allowed to request, and using the tool, the LLM might say something like, hey, I want to run this Python code, and then an application like Positron goes and runs it and hands the results back to the LLM.
Because at its core, an LLM alone is just text in and text out. It can write you Python code, but by itself it doesn't have any way to run that Python code. It's just text.
Again, the LLM is just taking text in and outputting more text. So when one AI product can, say, browse the internet for you and another one can't, even though it has the same access to the same underlying model, it's because they have different tools.
