In a recent interview, Nvidia CEO Jensen Huang acknowledged that current AI technology still falls short of providing reliable and trustworthy answers.
While artificial intelligence, especially large language models like ChatGPT, has made impressive strides, Huang believes that we are still several years away from achieving an AI system that can be fully trusted.
Huang emphasized the issue of “hallucinations” in AI, where the system provides false or fabricated responses. He pointed out that users should not have to second-guess whether an AIโs answer is accurate, a problem that persists in today’s models.
He stressed that more computational power will be required to improve the reliability of AI systems and ensure that answers can be trusted.
Despite the advancements in AI, Huang noted that pre-training AI models on large datasets is not enough to reach the desired level of trust. He compared it to completing a college degree, which is a significant milestone but not the end goal.
As companies continue to enhance AI technology, Huang believes that the journey toward trustworthy AI is still in its early stages, requiring further improvements in both computational power and model development.
