Content
@
0 reply
0 recast
0 reaction
Paul Berg
@prberg
Does OpenAI feed user conversations back into training future versions of GPT?
6 replies
16 recasts
30 reactions
mdali.eth
@mdali.eth
Most likely they do, similarly to how Tesla is training their autopilot on countless hours of human driving, including any active corrections by the driver when autopilot is engaged. Crowdsourced training is just too powerful. Ask Google about reCAPTCHA’s training.
0 reply
0 recast
1 reaction
rev
@rev
there's a way out 🤔 https://warpcast.com/rev/0xc999b88d
0 reply
0 recast
1 reaction
bitsofjeremy (.eth, .sol)
@bitsofjeremy
Yes. Go into the app and look at data controls. You can flip it on or off.
0 reply
0 recast
0 reaction
Chris
@chrisdom
https://frametra.in/f/gytnq6rumroxbuf8qik9qiw9
0 reply
0 recast
0 reaction
Dawn Kelly
@dawnkelly.eth
I just assume anything I put in there can be read by someone, somewhere. Machine or human? Idk. I just assume it’s never private
0 reply
0 recast
0 reaction
Streamm.tv
@retake
I think it’s on their tos..
1 reply
0 recast
0 reaction