This year, I've noticed how folks are using AI has changed.
Before, people were just looking up info, writing copy, and doing translations; now, more and more are starting to chat with AI about work, entrepreneurship, investing, and even some personal stuff they wouldn’t easily share.
Not long ago, a buddy of mine checked his health report and found some odd indicators. His first instinct wasn’t to ask a doctor; it was to fire up AI. But halfway through typing, he deleted it. He was worried that his health, income, and family info would get logged, analyzed, and possibly become data for future model training.
This made me realize that the biggest issue with AI might not be its intelligence but whether it’s worth trusting.
That’s why I’m keeping an eye on @OpenGradient .
Many projects are still grinding on model capabilities, while OpenGradient is more focused on privacy and verification. OpenGradient Chat separates “who you are” and “what you asked” as much as possible through encryption, relay networks, and TEE; at the same time, it employs a verifiable reasoning mechanism that allows the AI's results to be validated, rather than being a black box that can’t be checked.
Of course, whether this architecture can balance privacy, cost, and user experience will take time to verify.
But at least it’s discussing an increasingly important question: when AI becomes everyone’s long-term assistant, how do we actually trust it?
In the future, what determines AI's competitiveness might not just be intelligence, but trust.
#OPG $OPG
Before, people were just looking up info, writing copy, and doing translations; now, more and more are starting to chat with AI about work, entrepreneurship, investing, and even some personal stuff they wouldn’t easily share.
Not long ago, a buddy of mine checked his health report and found some odd indicators. His first instinct wasn’t to ask a doctor; it was to fire up AI. But halfway through typing, he deleted it. He was worried that his health, income, and family info would get logged, analyzed, and possibly become data for future model training.
This made me realize that the biggest issue with AI might not be its intelligence but whether it’s worth trusting.
That’s why I’m keeping an eye on @OpenGradient .
Many projects are still grinding on model capabilities, while OpenGradient is more focused on privacy and verification. OpenGradient Chat separates “who you are” and “what you asked” as much as possible through encryption, relay networks, and TEE; at the same time, it employs a verifiable reasoning mechanism that allows the AI's results to be validated, rather than being a black box that can’t be checked.
Of course, whether this architecture can balance privacy, cost, and user experience will take time to verify.
But at least it’s discussing an increasingly important question: when AI becomes everyone’s long-term assistant, how do we actually trust it?
In the future, what determines AI's competitiveness might not just be intelligence, but trust.
#OPG $OPG