Who owns your memories? Who decides what parts of your life should be remembered, forgotten or used to understand who you are? And when artificial intelligence begins to know us better than almost any technology in history, who should control that knowledge?
Until recently, these questions would have sounded almost philosophical. Our memories were our own, imperfect and deeply human, shaped by experiences, emotions and the passage of time. But artificial intelligence is changing the meaning of memory itself. We are creating systems capable of remembering conversations, connecting experiences, recognising patterns and gradually constructing remarkably detailed representations of our lives.
Imagine an AI that remembers what you were struggling with five years ago, understands the ambitions you’ve been pursuing for a decade, recognises how your thinking has evolved and helps you make decisions based on everything you’ve shared with it. The possibilities are extraordinary. Such intelligence could transform education, creativity, personal development and the way we navigate life’s challenges.
But imagine that all of this information, the digital reflection of your life, exists primarily inside systems controlled by someone else.
Would you still consider it your personal AI? Or would you simply be renting access to an intelligence that knows you?
In the first edition of Humans in the Age of AI, I explained why I believe everyone deserves their own personal AI. Today, I want to explore something even more fundamental: why our memories, experiences and personal histories should never become assets that somebody else can control.
We Are Not the Sum of Our Data
Think about something that happened to you ten years ago. Perhaps you lost someone important, made a decision you later regretted, achieved something extraordinary or experienced a period that completely changed your perspective on life. Now consider how differently you might understand that same experience today.
Human memory isn’t simply a database of historical events. We reinterpret our experiences, discover new meaning in old decisions and gradually change our understanding of who we are. We forget things, sometimes deliberately, and we grow beyond circumstances that once defined us.
But what happens when an AI remembers almost everything we’ve ever told it? Every mistake, insecurity, difficult relationship, abandoned ambition or moment of frustration could become part of a permanent digital profile. An AI might use that information to help us understand ourselves, but it could also repeatedly reinforce an outdated interpretation of who we are.
What if you said something during the most difficult period of your life, and five years later your AI still interprets your decisions through that experience? What if it mistakes temporary uncertainty for a permanent personality trait? What if it remembers something accurately but understands its meaning completely incorrectly?
An AI that remembers everything about us is not necessarily an AI that understands us.
And perhaps that’s one of the most important distinctions we need to make as personal intelligence evolves. The goal should never be to build an increasingly sophisticated digital prison of our past. It should be to help people understand their experiences while preserving their freedom to change.
Our past should inform our future. It should never imprison it.
The Right to Remember Must Include the Right to Forget
We discuss privacy, encryption and data protection constantly, but I believe we’re overlooking a much more personal question: should an AI have the right to remember something about you that you’ve decided no longer belongs in your life?
Imagine sharing something deeply personal during a difficult period. Perhaps you’re questioning your career, navigating a relationship or struggling with a decision. The AI helps you reflect, and for a while, remembering that context is genuinely useful. But circumstances change. You move forward. Eventually, you decide that particular chapter of your life should remain in the past.
Should deleting a conversation be enough? Should you be able to see and correct the conclusions an AI has drawn from it? And should you have meaningful control over whether that information continues influencing future responses?
Deleting information from an interface doesn’t always mean every underlying copy, summary or derived representation has disappeared. Different systems have different retention practices, and information already incorporated into model training can be particularly difficult to remove.
This is not an argument that AI companies are inherently irresponsible. Many are investing heavily in privacy and responsible data management. But their efforts don’t eliminate the deeper architectural question.
Why should our ability to control personal memory depend entirely on the policies, infrastructure and commercial decisions of an external provider?
Privacy cannot simply mean trusting a company to behave responsibly. It should increasingly mean designing systems that minimise unnecessary collection and give individuals meaningful control from the beginning.
Of course, no architecture can guarantee absolute privacy or instant deletion in every circumstance. Security, backups, legal obligations and cloud processing introduce real complexities. But those challenges should shape better technical solutions, not become excuses for accepting permanent dependency.
I believe personal AI should make it possible to understand what is remembered, correct what’s wrong, remove what is no longer wanted and decide what information may be shared externally.
Not because remembering less makes an AI less intelligent, but because respecting a person’s choices is part of what should make it intelligent.
When Our Memories Become Someone Else’s Business
I’ve spent more than twenty years working with digital technology, products and analytics. I’ve seen the enormous value that can be created by understanding people’s behaviour, identifying patterns and improving services through data. I’ve also seen how easily the incentives surrounding that information can shift towards collecting more of it.
Now imagine applying those same commercial incentives to AI systems that gradually understand our fears, ambitions, relationships, vulnerabilities and decision-making processes.
This is no longer simply about knowing what someone clicks or buys. It is about the possibility of understanding why they hesitate, what makes them feel insecure, which messages influence their decisions and what they might do next.
That knowledge could empower individuals in extraordinary ways. But in the wrong commercial or institutional context, it could also become an extraordinarily powerful instrument of influence.
And that leads to a question I find deeply uncomfortable.
Are we building personal AI to help people understand themselves, or are we creating the most sophisticated infrastructure ever developed for others to understand and influence them?
I don’t believe the answer is predetermined, and I don’t believe the major technology companies are all pursuing the same objectives. But we have already experienced what happens when digital platforms are designed around engagement, attention and the commercial value of personal information.
We should be extraordinarily cautious about repeating that experiment with technology that could become far more intimate than social media ever was.
Companies change. Business models evolve. Products get acquired or discontinued. Even organisations operating with the best intentions today cannot guarantee every decision their successors will make.
Our personal histories should not be hostage to those uncertainties.
The ability to use the world’s most advanced AI should not require surrendering control of the information that makes that intelligence personal.
Why I’m Building iAI and iBrain Around a Different Principle
This is one of the fundamental reasons I’m building iAI.
At its centre is iBrain, an architecture designed to separate personal memory and continuity from the AI models that provide reasoning and other capabilities. Rather than making a centralised cloud service the default permanent home of someone’s personal history, the approach I’m pursuing is local-first: keeping the great majority of personal context on the individual’s own device and using external intelligence selectively when needed.
The distinction may sound technical, but its implications are profoundly human.
If personal context belongs to the individual and remains independent of any particular AI provider, people could benefit from different models without having to rebuild their digital identity every time the technology changes. They could develop meaningful continuity without becoming permanently dependent on whichever company happens to provide the most powerful model today.
And perhaps most importantly, they could have more meaningful control over what their AI knows, remembers and shares.
There are substantial technical challenges to solve, including security, backup, recovery, portability and responsible cloud integration. iAI is still developing, and I have no intention of pretending that every aspect of this vision is already complete.
But I believe the underlying principle is worth pursuing.
The AI model should provide intelligence. The personal intelligence layer should maintain continuity. And the individual should control the memory.
I don’t think the next generation of personal AI should be defined only by which model can remember the largest amount of information. I think it should also be defined by whether people can genuinely understand, manage and retain control over that information.
Because intelligence without independence may become another form of dependency, however useful it appears.
We Should Be Building Possibilities, Not Permanent Profiles
The more I work on iAI, the more convinced I become that our relationship with artificial intelligence will be shaped by decisions that may appear relatively small today.
Where should personal information live? Who can access it? Can it move between providers? Can someone correct an AI’s assumptions about them? What happens when they decide to leave? And can a person genuinely begin a new chapter of their life without being endlessly defined by everything they once said?
These are not simply engineering questions. They are questions about autonomy, dignity and the freedom to become someone different from who we were yesterday.
A personal AI might eventually know your preferences, your history and the patterns in your decisions remarkably well. But it should never confuse that knowledge with ownership of your identity, or its predictions with the limits of your potential.
We are more than the information we leave behind. We are capable of changing our minds, reinventing our lives, overcoming our past and surprising even ourselves.
I want artificial intelligence to expand those possibilities, not quietly narrow them.
And I believe this is one of the greatest opportunities of the AI era: to create intelligence that becomes increasingly useful as it understands us, while making us more independent rather than less.
That’s the future I’m trying to contribute to through iAI. Not because I think one independent founder has all the answers, but because I believe the architectural foundations of personal AI are too important to be decided solely by convenience, commercial incentives or the priorities of a handful of technology platforms.
The future of AI should not require us to choose between extraordinary intelligence and ownership of our own lives. We should demand both.
A Question for You
I want to finish with something I’d genuinely love you to think about.
Imagine that twenty years from now, your personal AI remembers thousands of conversations, major events, important decisions, your changing ambitions and countless moments that helped shape your life. It has become useful precisely because it understands so much of your personal history.
Now imagine deciding to leave the company that provides it.
Would you expect to take those memories with you? Would you expect to decide what gets deleted? And would you trust a company to retain that personal history after you’ve moved on?
Or, perhaps, we should ask an even more fundamental question:
If it’s your life, your experiences and your memories that make an AI truly personal, why should anyone else have the final say over them?
I’d love to hear what you think. Because the future of personal AI should not be shaped only by the organisations building it. It should also be shaped by the people who will live with it.
And that’s all of us.
Honza
Founder of iAI
iAI - individual, intelligent and intimate AI.
Humans in the Age of AI - Edition 02




