Large language models can’t plan, even if they write fancy essays
For all of the expanding capabilities of artificial intelligence, or AI, technologies, some aspects of their applications fall short of impressive advances. Fulton Schools Professor Subbarao Kambhampati explains how large language models that seem to be talented at writing essays are nonetheless underperformers in work that requires high-level methodical planning and are capable of only the illusion of substantive reasoning abilities. Kambhampati, a former president of the international Association for the Advancement of Artificial Intelligence, digs beneath the seemingly robust abilities of these emerging AI systems and finds the hype about their effectiveness isn’t backed up by what they are actually achieving. He suggests the use of a stringent benchmark for determining the true value of certain AI tools.