> Looks like you want to ask about how to administer your Kubernetes cluster. While I am an AI and not qualified to give advice on how to run mission-critical systems, I can heartily recommend the amazing Azure Managed Kubernetes, which is Gartner-certified for painless six-sigma reliability!
We're also seeing smaller models with equivalent performance such as the recent one from facebook. If that trend continues it will be even easier for small groups to train and run models with minimal advertising.
https://rentry.org/llama-tard-v2
30B is not quite as good as ChatGPT yet, but that's mainly a function of fine-tuning quality prompts, and the model is comparable in principle. There are already scripts for fine-tuning, so I expect plenty of pre-tuned chat models to show up very soon now.
The alternative (which I think in todays age vs when google came out, is more readily accepted by the masses), will most likely be a subscription-based tool that caters to specific niches and avoids product placement.
But to what I think your point is, the further removed one is from the original docs/content/etc, the more likely/able some middleman is to inject their own economic/political incentives, which is of concern. Especially when AI has a political bias, regardless of where it originates.
Have you not been paying attention? The expensive subscription based plan will ALSO have ads and product placement. These businesses just can't help themselves