AI isn't really that smart yet, Apple researchers warn
Apple’s research reveals that large language models are highly fragile, failing when inputs change slightly. This instability means they cannot serve as fully autonomous decision-makers but remain useful as human adjuncts. For open data, this highlights the critical need for robust, high-quality datasets to stabilize AI reliability and prevent erratic outputs.
Source: computerworld.comPublished on 2024-10-17
Related news
- India has great advantage in building new AI applications on top of LLMs: Accel partner Shekhar Kirani
- Tales aims to revolutionise Gaming AI to enable users to build and play any game
- NYT Issues Cease and Desist notice to Perplexity for AI Scraping
- SKYSCENES Boosts Autonomous Flight Safety with AI
- The Internet Archive case: implications for India's copyright landscape