DailyPost 3020
HOW DOES AGENTIC AI WORK?
Even for the ones who got initiated into Artificial Intelligence (AI) even as late as 30th Nov. 2022, when ChatGPT was launched, would agree to the current trajectory of AI; that is from now famous Generative AI to Agentic AI. Being autonomous in nature, Agentic AI is quite different from generative AI, as it primarily focuses on decision-making and actions rather than creating content like text and images and varieties of such other correlated tasks. This transformation of the AI landscape can completely change every aspect of human existence in a manner no one would have imagined even six months back.
Agentic AI is our future as we stand at this moment in AI history. Given this reality the ever inquisitive minds of the AI age would certainly like to know, as to how does this happen? What is the process and mechanism one would like to know? Agentic AI uses a structured process that allows it to autonomously achieve its goals. The greatest trait is that it keeps adapting to the new information and improves over time. A breakdown of the process will clarify how it works in the real world. It starts with goal setting. The process begins with defining a clear objective or the problem addressed to Agentic AI to solve, perform, execute or resolve.
The next step in the main kernel of its functioning typical of this tool is data garnering and processing. Having established the goal, agentic AI processes data from various data sources, including cloud platforms and IoT devices. Then the agentic AI transitions to the planning and problem-solving stage. This multifarious data is analysed real time, “the agentic AI system gains insights needed to make an informed decision.” The next stage is planning and problem-solving. Advanced algorithms are put in use by the agentic AI to develop a multi-step plan to complete its goal. The beauty of this AI tool is that it can identify potential obstacles and “adjust its approach to overcome any challenge as they arise.”
Next is the execution stage on what has been planned in the earlier stage. This whole exercise is conducted without human intervention. All actions are goal oriented and designed to deliver measurable outcomes. There is no requirement of constant oversight. The completion of the task successfully is not the end of the process. It goes on to evaluate its performance by analysing results against the original goal. This feedback is used to refine its algorithms and improve future performance. This ensures continuous learning and efficiency. This is the marvel machine, in its nascent state, being developed to function as a general purpose technology.
AGENTIC AI IS LEAVING SCI-FI BEHIND, AND IT’S REAL!
Sanjay Sahay