"According to a definition by IBM, agentic AI consists of software agents that mimic human decision-making to solve problems in real time, and this builds on generative AI techniques by using large language models (LLMs) to function in dynamic environments. But while the industry hype machine pushes agentic AI as the next big thing, potential adopters should be wary, as"
"the paper, "Hallucination Stations: On Some Basic Limitations of Transformer-Based Language Models" [PDF] argues that LLMs are incapable of carrying out computational and agentic tasks beyond a certain complexity level, above which they will deliver incorrect responses. The paper uses mathematical reasoning to show that if a prompt to an LLM specifies a computational task whose complexity is higher than that of the LLM's own core operations, then the LLM will in general respond incorrectly."
Agentic AI uses LLMs as software agents that mimic human decision-making and operate in dynamic environments. LLMs possess a complexity ceiling: when asked to perform computational tasks whose complexity exceeds their core operations, they generally produce incorrect results. Verification of another agent's solution can be more complex than the original task and therefore may also fail. Mathematical reasoning underlies these complexity-based limitations. These constraints threaten applications that automate real-world effects, including software generation, financial transactions, and industrial control, so extreme caution is warranted when deploying agentic systems for high-complexity tasks.
Read at Theregister
Unable to calculate read time
Collection
[
|
...
]