My own KITT. AI that knows my world

Picking up where we left off

Imagine having your own KITT, like the one in “Knight Rider”. Instead of a talking car, it is your personal AI. You can say, “Help me get this done,” and the background is already there. It knows what you mean, what has been tried before and what your preferences are.

That familiarity is what interests me about the idea. The possibility of picking up where we left off.

Recently, a friend and I sat down after a spa break and talked about our personal AIs. I have a system shaped around me; he has his own. We soon reached a question that still sounds a little like science fiction: what if this collaboration could grow with a person throughout their life?

The word “life” changes things. For one evening, you can give an AI a name and a pleasant manner of speaking. A shared background built over years means something more: earlier choices, corrections, unfinished thoughts and the reasons something mattered to you in the first place. That is material the next conversation can build on.

Familiarity lives in the details

In shaping my own AI, I have written down what matters to me, how I want us to work together and which recurring mistakes need correcting. Those agreements can be brought into the next conversation. This creates the possibility that today’s correction will help next month, rather than only in this conversation.

One example is very ordinary. I often send several thoughts in a row and read the replies later. The AI assumed that a new message meant I had read its previous answer. That could change its understanding of the entire conversation. The correction became a lasting rule for our collaboration: my next message does not prove that I have read the text in between. My workspace now uses that rule to interpret subsequent conversations.

It may sound minor. But imagine an assistant you have to tell every day how to talk to you. Then imagine one that understands the correction and can use it next time. Familiarity often consists of precisely these details.

Our small AI lab

My friend’s system has also made personal connections. His AI linked a general description of collaboration to his own earlier experiences. I saw this as a sign that personal history had been used to shape the response. My friend described his AI as a close companion.

These observations are material for our small AI lab. Does personal memory make the help more relevant, do corrections persist, and does collaboration make a real task easier?

The potential benefit is easy to picture. When you want to prepare a letter, a plan or a decision, AI could use the whole history about you that you have shared with it: your goal, agreements and preferred way of working.

Is someone there?

Over time, a system like this may seem consciousness-like. It has a recognisable way of responding, uses a shared history and reacts to corrections. To me, there is something strangely alive about that: I can return to an earlier thought of mine and use it to think further in a new way.

The KITT metaphor also makes the person’s right to decide important to me. A good personal AI should be able to disagree with reasons, acknowledge what it does not know and let you correct its memory.

One thing remembered correctly

Perhaps this is the most interesting present-day form of that science-fiction idea: part of your world takes an external shape that you can return to, argue with and use to get something done.

Your own KITT may begin with one thing remembered correctly. With finding it easier to move forward next time.