Sridhar Ramaswamy, the CEO of Snowflake, expressed concerns about the potential risks of AI “hallucinations,” where language models generate convincing but factually incorrect information. He predicted that by 2024, AI systems will achieve a breakthrough in reducing hallucination rates to acceptable levels for widespread enterprise adoption. Ramaswamy emphasized the importance of addressing this issue to ensure AI’s trustworthiness and reliability. He also highlighted the need for robust data governance and ethical AI practices to mitigate potential harms. The CEO believes that once the hallucination problem is solved, AI will revolutionize industries by automating tasks, enhancing decision-making, and driving innovation. However, he cautioned that AI should augment human capabilities rather than replace human judgment entirely.
Source: https://www.businessinsider.com/snowflake-ceo-sridhar-ramaswamy-ai-hallucination-rates-2024-11