Your Own Scott Galloway

A couple of weeks back I wrote a story on using small LLMs. Well, I wanted to revisit that post.

Scott Galloway has trained an AI chatbot. This chatbot is based on his Podcasts and blog posts he wrote. I find this exciting as I explained in my Tiny Stories post with the help of AI. Scott Galloway validated my premise with his Profg.ai.

This chatbot will give you access to answers Scott Galloway would give. Though this is an interesting experiment I think this can be taken further. It is just the beginning of what these products can do. I see them as more for entertainment uses than they are for practical everyday advice.

You will not get personalized advice from these chatbots but if [insert influencer, business guru] had a bot that could lead to new lines of business and new ways for influencers to monetize their work.

How far this will go is TBD. However, as language models become better at mimicking use the line between you and the chatbot will blur.

I am convinced that there is a lot of misunderstanding of how much of a change using a smaller language model is.

Get this, you can now create a chatbot utilizing the best data possible. As it now stands chatbots are built hoovering up an ocean of data to build their language models. What if you didn’t need to do that, but instead you could choose the data you are using and you could choose how much of that data you want touse? Your language model could then be small, and they would be easier to use and maintain.

What the lesson of Tiny Stories teaches us is that building a language model could become something many more could do. We could do this because there would not be a need for large data servers to hold and manage the data. One such model already exists, Orca. and I think there will be more to come.