im not saying trust 'big AI'. im looking forward towards the future when we're all running open source models on our pocket computers.
Discussion
If it needs to be connected to the internet to be helpful, how do you keep it from getting compromised?
it wouldn't need to be if the model lives on your phone
How so? Let’s say I tell it to book tickets to Ironman 7 while I get all the kids loaded in the van. Are you saying it could just control your phone, open safari and book tickets that way?
that's one method that some have done. that's what the rabbit r1 essentially did a couple years ago. it ran a VM and actually opened apps and used them. that could happen on the user's phone in the future. where AI just interacts with your apps, but that may not be as efficient.