News Summary

• Researchers at the University of Colorado at Boulder challenged large language models to solve sudoku.
• An easier 6x6 puzzle was often beyond the capabilities of an LLM without outside help.
• A more important finding came when the models were asked to show their work.
• For the most part, they couldn't.
• sometimes they lied.
• Sometimes they explained things in ways that made no sense.
• If gen AI tools can't explain their decisions accurately or transparently, that should cause us to be cautious.