AI Chat Preference

[gtyellowjacket]

Read something the other day saying that all of the models were trained on the same public domain data (wikipedia, reddit, stack overflow) and will become commoditized in the near future. What will differentiate an AI model in the future is private data - eg: Westlaw or LexisNexis for legal. Years of historical payment data for a bank/financial institution.

I think this is right. But the thing to remember is that folks like westlaw or lexus nexus are not creating their own gemini or openai. They're using a general model like Gemini and openai or similar to access their private data.

That is really the way copilot works for the most part although there are much more sophisticated and better tools out there. But accessing private data is probably going to give companies an advantage and the general model that provides the best results at the lowest price will win as the backend tool of choice.