NVIDIA DIRECTOR SAYS AI SAFETY DOES NOT NEED TO SLOW DOWN THE FRONTIERAli Golshan, Senior Director of AI Software at Nvi | Hanami
NVIDIA DIRECTOR SAYS AI SAFETY DOES NOT NEED TO SLOW DOWN THE FRONTIER
Ali Golshan, Senior Director of AI Software at Nvidia @NVIDIAAI, said on my exclusive interview with him
- Security has always sped things up, not slowed them down: "We did more and did it faster on the web when SSL became a default, when encryption became a default." - Agents fail or drift off-task specifically because they don't have enough trust to work with real context: "We are lacking trust and privacy, because we can't give them all the things we want, otherwise we are not sure where they could go." - His analogy: hiring a contractor and saying "fix the leak" blind is a coin flip. Walk them to the actual crack, and "it is much more likely that the solution will actually end up being the right one."
The bet behind Nvidia's OpenShell platform: companies that build trust and governance into their agents first will be able to hand them more sensitive data and more autonomy, which means faster, not slower, progress.