DailyPost 3046
AGENT AI FOR GOVERNMENT
IBM Watson brought AI into public knowledge, was unveiled in 2011. It was developed on the game show Jeopardy, where it defeated human champions. Later it evolved into a powerful AI platform used for various applications. The story of Demis Hassbis and DeepMind is all too well known. The Alpha series has done yeomen service in the development of AI, when there were not many takers. Recently Demis Hassabis shared the Nobel Prize for his work on AlphaFold. OpenAI founded in 2015 has now been known for the groundbreaking AI model GPT. This led to the creation of ChatGPT, released in public domain on 30th of Nov. 2022, took the world by storm.
This was generative AI for use of one and all, primarily being trained on internet / public domain data. It has led to its generic use in a variety of ways. Governments have remained bystanders to the progress of AI, inclusive of handling its ethical issues and the likely regulatory issues which have already started cropping up. To have a tool which could be used for specific organisation, purpose or task, was slowly turning out to be the need of the hour. Use case for any technology is its real application. To cater to this requirement, the present stage of AI, called the Agentic AI was born. The AI agent would deliver the exact requirements.
2025 has been proclaimed as the year of Agent AI. Still in a very nascent stage, the corporates globally are getting ready to hop onto this bandwagon. The governments are yet to get a hang of it. Most of the developing nations still have lots of crevices in the digital ecosystem for a variety of reasons. Hence, the intelligence and seamlessness that needs to be generated out of this system is hugely truncated. The databases are in silos, the integrity of data is also in question. Its reconciliation is a big task. The workflows are broken and it cannot be worked upon. Agent AI can turn out to be a one stop solution to most issues to start with and can improve iteratively, in a very fast manner.
The turnaround time for Agent AI is negligible, if it is developed in the right manner. “An AI Agent is an autonomous system that can perceive its environment, take actions to achieve goals, and potentially improve its performance through learning and acquiring knowledge.” What more do you want a tool to do? Domain knowledge to make it happen, has to come from governance practitioners. The right LLM or SLM has to be selected for the purpose. All business logic can be built into it. An Agent AI for each of the tasks / enterprise level and each Agent AI performing its task and all Agents stitched into a system can do wonders. Time need not be lost, models of proven worth are already there, the whole world is experimenting, governments cannot be an exception, ending in putting up truncated systems when forced to later.
AGENT AI CAN WIPE OUT LARGE AMOUNT OF GOVERNMENT DISCREPANCIES TO USHER IN A DYNAMIC AND REAL TIME DATA DRIVEN SYSTEM.
Sanjay Sahay