OpenAI is now using constrained decoding, a technique that may help  keep artificial intelligence technology from making mistakes by helping to ensure there is consistent data generation when it is attempting  provide complete and accurate information. In one recent attempt at clear reasoning, things got murky in relation to Jagged Intelligence, a term used to define the struggle that AI’s Large Language Models, or LLMs, have in dealing with dumb problems. Subbarao Kambhampati, a professor in the School of Computing and Augmented Intelligence, part of the Fulton Schools, explains that LLMs are impressive tools but there remain challenges that need to be addressed regarding the efficacy of LLM-based AI Agents.