NVIDIA CEO Jensen Huang said that today's artificial intelligence (AI) doesn't always provide accurate answers and that the industry is still in its infancy, write Business Insider.
"Today the answers we have are not the best we can give", Huang said.
NVIDIA's CEO noted that people don't need to second-guess the AI's answer by wondering whether it's "hallucinating or not" or "intelligent or not."
"We've got to get to the point where the answers you get—you'll pretty much trust— he said — so I think we'll be able to do that in a few years, and in the meantime, we have to keep increasing our calculations."
Large language models like ChatGPT have advanced exponentially over the past few years in their ability to answer complex questions, but they still have limitations. Some AI companies are also struggling with how to promote LLMs without relying solely on large amounts of data and an already limited resource.
In an interview on Saturday, Huang said that pre-training, or training a model on a large, diverse data set before developing it for a particular task, would not be enough.
"Preliminary preparation — simply taking all the data in the world and automatically discovering knowledge from it is not enough for preliminary preparation." Jensen Huang said.
