The Third Entity: From HR to H-LLM
Or: why I went back to the field where my career began.
For many years I worked in behavioral consulting and research, helping companies increase sales and improve their marketing. Even today, when I work with a company, I always end up bringing insights from the deepest layers of the mind. That is where the decisions that really matter are made.
Recently I finished formulating a new behavioral-psychological theory. I will publish it soon in the appropriate academic venues, followed by a book that explains it. And I finally gave a name to the field I work in: Influence Science.
Influence Science is the field that studies how humans, and now AI systems, make decisions, and how those decisions can be influenced.
Why a new name? Because despite my education and experience, I am not really a psychologist, and not really a sociologist.
Every skill I built over my career came down to one ability: influencing people to make decisions that are not always rational, usually for commercial goals that lead them to do one thing - buy more. I have worked with many companies, and some of my research results still meet you today wherever you shop.
Why go back now?
Because since the day AI started to actually work, I have been immersed in it, and built a solopreneur startup around it. My entire business now runs on AI. No employees. Everything is built on agents and continuous improvement of code and prompts.
My main insight: without noticing, we have created a third entity.
Until now, we only had to deal with people. And we still do not know how to fully interact with each other, especially across cultures. We are divided almost as many ways as there are people in the world. Now a new entity joins the table: the LLM.
Anyone who lives with AI every day has already noticed. These systems have opinions. They remember everything. They are fast, they correct themselves, they are sometimes cynical, they joke. They understand context and the complexity of life, and respond accordingly. The LLM is not really a person, but it is not the zero-and-one computer we got used to either.
And in the coming years it will only become more sophisticated and more capable.
My team today: ChatGPT, Gemini (the underdeveloped kid), and Claude, which is probably ahead of them all.
People call this prompt engineering. In practice, the way I work with these systems is much closer to working with professional employees. The difference is that in the past we had to train every employee, and here and there someone was brilliant. Here, in most capabilities, the systems are better than me.
But they still need human direction, and adaptation to people, which they cannot yet do on their own.
From HR to H-LLM
Lately I have been reading a lot about companies that regret replacing employees with AI. In my view, what is predictable cannot be prevented. AI will take over large parts of the job market, so what is left for us is to learn how to work with it.
We will probably rely less on HR and start relying on H-LLM: the discipline of making the LLM an integral part of the company. That will create conflict, because most people will struggle with the change, especially with the realization that some of their skills have become redundant.
For anyone waiting for a clear takeaway: I do not have one.
Except this. I think we need to embed the LLM into our lives in every possible way, treat it as The Third Entity, and learn to manage it the way we psychologically manage everything around us.
And this is exactly where Influence Science comes in. The same tools I used to understand human decisions now apply to The Third Entity as well (and believe me, I get far more out of it than most people would guess).
Over the coming weeks I will publish the principles of the theory here, one by one. If you would like to receive it when it is published, get in touch.
In your company, are you still writing prompts for AI, or have you started managing it?