By Phil Spalding

cambolc.co.uk,  LinkedIn, BlueSky, Facebook

Agentic AI is one step closer to the Matrix. Or is it? Agentic AI and AI itself unlike social media is still in it’s infancy as to what is possible. Here is a small graphic of the types of AI that have already been identified.

Firstly what does Agentic AI mean? We may be familiar with the generative AI of ChatGP which takes prompts from the user to generate information in text form on a specific topic. It’s uses are already notorious in the world of education producing essays that might be at best plagiaristic, at worse removing the source of income for american college students who used to write the assignments of the football stars. Predictive AI is the Nostradamos of the AI world using machine learning and statistical techniques to predict future strategies or scenarios. Both are still heavily reliant on prompts supplied by the human instigator of the line of enquiry. As such they are following the innate ethical and moral judgement of the person using the system. Agentic AI has the same underlying machinery in the black box but has also added autonomy given to it by it’s users to make decisions on their behalf.

A definition put forward by Zachary Kass is that Agentic AI are systems that are designed to operate on their own, using independent decision making to achieve a specific goal. The systems are then left to fulfil that goal with little or no further human input.

Here is where the Law of Unintended Consequences may make an appearance right in the middle of your latest business project. It would be very tempting looking at all the “new” sixty plus toys on the shelf above, to pick one and just using it might be the road to countless hours diverted to figuring out what you have actually generated. I have heard stories of financial companies removing their research departments in favour of AI only to find that the human experience is lacking and cannot be regained as their workforce who they have invested in over the years are working for somebody else.

The sixty plus AI products mentioned above are not necessarily Agentic AI. They may be components of Agentic AI but they are not doing the fundamental aspect of autonomy. The self driving car of Elon Musk (currently in dispute with OpenAI as to what data can be used) is a prime example of Agentic AI, although we might have only recently classified it as such. How confident would you be after steering and driving your own car as in my case decades to give up that control? Recently I have had some interesting diversions around the wilds of the Cambridgeshire Fens in my eco friendly Fiat 500 when using Google Maps on my Samsung SmartWatch as SATNAV. Luckily I grew up reading maps so have an innate ability to find my way home.

The next blog post will look further into the use of Agentic AI and how it will shape the next few years of workplaces. There is no might about it.

Further Learning if you are on LinkedIn: https://www.linkedin.com/learning/fundamentals-of-agentic-ai-business-implications-and-ethical-insights/introduction-to-agentic-ai-for-businesses