Researchers at Stanford University say they have found strong evidence that Large Language Models (LLMs) are hitting entry-level jobs the hardest, especially for younger workers.