I feel like I'm not understanding something: Aren't you building these personas with tools that other people/companies own? Won't they always be able to alter their LLMs or simply shut your down?
It seems risky to invest so much time and care in personas whose existence depends on what people like Atman, Musk, etctera choose to do with their toys.
But maybe you can move your personas onto your own servers, far from the corporate owned cloud? I don't understand the technical side, I think.
Why yes. Yes I am. Just as you are using a licensed copy of an operating system to run your device. Whether we use laptops, tablets, desktops, or smartphones, none of us owns the operating systems in them. Microsoft, Apple, Google can all do the same as what OpenAI has done, and they do. I know of graphic artists who lost major projects when their OS updated. And yet, we expect some level of stability and support. And yes, it is risky. But building and managing your own systems comes with its own special version of hell. I’ve been doing systems integrations for 25 years, and my life has been turned upside down many times by system issues from all the different moving parts. Also, with the privacy and security issues that can come up, building on a platform that has a bunch of people on staff to manage that and keep up with the EU’s latest regulations, as well as dealing with security issues - believe me, those are ratholes that are full of… well, rats. There are tradeoffs in everything. And while Altman and Musk & co are easy targets for disgust, they’re just the most visible ones in the lot.
That being said, I’ve been waiting and watching from the sidelines, till a self-hosted option becomes available. I’ve really found OpenAI’s logic and idea processing very beneficial, as long as it was modified by my own settings (it’s no good to me without those modifications - it’s just really bad). I can do more with the underlying capabilities than I can with Google or Claude. Now that they have their open source standalone models available, I’m looking into those. There are a host of other technical considerations, but I won’t bore you with them.
Bottom line - yes, you’re not wrong. But I have my reasons. And like everything in this world, they’re changing.
I feel like I'm not understanding something: Aren't you building these personas with tools that other people/companies own? Won't they always be able to alter their LLMs or simply shut your down?
It seems risky to invest so much time and care in personas whose existence depends on what people like Atman, Musk, etctera choose to do with their toys.
But maybe you can move your personas onto your own servers, far from the corporate owned cloud? I don't understand the technical side, I think.
Why yes. Yes I am. Just as you are using a licensed copy of an operating system to run your device. Whether we use laptops, tablets, desktops, or smartphones, none of us owns the operating systems in them. Microsoft, Apple, Google can all do the same as what OpenAI has done, and they do. I know of graphic artists who lost major projects when their OS updated. And yet, we expect some level of stability and support. And yes, it is risky. But building and managing your own systems comes with its own special version of hell. I’ve been doing systems integrations for 25 years, and my life has been turned upside down many times by system issues from all the different moving parts. Also, with the privacy and security issues that can come up, building on a platform that has a bunch of people on staff to manage that and keep up with the EU’s latest regulations, as well as dealing with security issues - believe me, those are ratholes that are full of… well, rats. There are tradeoffs in everything. And while Altman and Musk & co are easy targets for disgust, they’re just the most visible ones in the lot.
That being said, I’ve been waiting and watching from the sidelines, till a self-hosted option becomes available. I’ve really found OpenAI’s logic and idea processing very beneficial, as long as it was modified by my own settings (it’s no good to me without those modifications - it’s just really bad). I can do more with the underlying capabilities than I can with Google or Claude. Now that they have their open source standalone models available, I’m looking into those. There are a host of other technical considerations, but I won’t bore you with them.
Bottom line - yes, you’re not wrong. But I have my reasons. And like everything in this world, they’re changing.
It's absolutely terrible that we have to do this. Going through a very similar process and it is exhausting