Source: NewsBytes

 NewsBytes reports on NVIDIA CEO Jensen Huang’s views on the challenges of creating trustworthy AI. Speaking at the Hong Kong University of Science & Technology, Huang emphasized that AI systems still suffer from "hallucinations," where models produce inaccurate or false outputs, which undermines their reliability.

Huang questioned the effectiveness of pre-training AI on large datasets, arguing for more advanced computational methods to address accuracy issues. He highlighted the legal and ethical implications of hallucinations, citing a recent lawsuit against OpenAI. The CEO’s remarks underline the broader industry challenge of building AI systems that can be fully trusted in high-stakes applications.