AI Workers at Google and Amazon Warn Families to Avoid Generative AI as Chatbot Error Rates Double to 35%
Summary
AI workers at Google and Amazon are warning their own families to avoid generative AI tools as a new audit reveals chatbot error rates have nearly doubled from 18% to 35% in just months, with employees citing rushed training and inadequate quality controls at major tech companies.
Key Points
- AI workers who train and evaluate models for companies like Google and Amazon are warning their friends and family to avoid using generative AI tools, citing concerns about accuracy and safety
- These workers report that companies prioritize speed over quality, providing inadequate training and unrealistic time limits while expecting them to evaluate complex topics including medical and ethical questions without proper expertise
- A NewsGuard audit reveals that chatbots' likelihood of repeating false information nearly doubled from 18% to 35% between August 2024 and 2025, while their non-response rates dropped to zero percent