The Last Choices
I am 94 years old. I am telling this story with some help. My hands aren't reliable enough to type anymore. The system understands my voice even when others sometimes have trouble. When I forget a name, it can usually find it. When I lose the thread of a story, it can help me return to it. That would have seemed miraculous when I was young. A great deal of my life would.
The world became strange
Cars eventually drove themselves better than we did. Many diseases became preventable. Others became treatable. AI tutors gave children access to education once available only to the fortunate. Translation became almost invisible. People routinely worked with colleagues they never physically met. Robots performed much of the dangerous physical work humans once did. Artificial intelligence became better than people at an astonishing number of intellectual tasks. Some jobs disappeared. Entire industries appeared. People lived longer. We also created new problems. It became difficult to know whether some things we saw or heard were real. People could live inside information worlds designed specifically for them.
Governments and companies gained capabilities to understand and influence human behavior that frightened me. Some people surrendered decisions to AI because it was easier. Others rejected technology almost entirely. Most of us lived somewhere in between. The future didn't become heaven. It didn't become hell. It became life. Complicated, astonishing human life.
HOS knew a great deal about mine
I had used it, in one form or another, for almost 60 years. It knew the decisions I had made. The things I once said mattered. The goals I abandoned. The mistakes I repeated. The people I loved. The person I had once hoped to become. That history became useful. It also required trust. Over the decades, I was repeatedly asked what could be retained, what could be used and what should disappear. I changed my mind more than once. The fact that something could help me did not automatically give anyone the right to keep it. That mattered more as technology became more powerful.
At 94, my problem wasn't creating more opportunities
I had fewer. My husband had died. Most of my closest friends were gone. I no longer drove. I needed help getting dressed. Some days I was sharp. Other days were foggy. My children worried about me. They wanted me safe. I understood. I also wanted to remain myself. Those objectives sometimes conflicted.
My daughter wanted me to move
She wanted me closer to her. It made sense. The predictive systems agreed. Lower health risk. Faster emergency response. More family contact. Better access to care. Every measurable variable favored moving. I didn't want to go. My daughter asked HOS to help convince me. It wouldn't. It helped us talk instead. Why did the house matter? Was I afraid of change? Was I underestimating the risks? Was my daughter overestimating them? What would make staying possible? What would make moving acceptable? Eventually I moved. Not because the system decided. Because I did. The difference mattered to me.
Then came a harder question
My cognitive abilities began declining. There were new treatments that might slow the decline. One carried significant risks. My children wanted me to try it. I wasn't sure. By then HOS had decades of my own reflections about health, independence, family and end-of-life care. It could remind me of what I had previously said. But it did something important. It didn't treat the woman I had been at 70 as having authority over the woman I was at 94. Past me could inform present me. She couldn't replace me. I chose a limited treatment. Later I stopped.
I began reviewing my life
Not because HOS told me to. Because I wanted to. It could reconstruct things I had forgotten. Photographs. Letters. Conversations I had chosen to preserve. Places. Decisions. There were moments when I barely recognized the person I had been. She was so certain. I wanted to tell her: “You don't know nearly as much as you think.” Then I realized my 110-year-old self might say the same thing to me. If I had one.
There were failures I never fixed
This surprised my grandchildren. They thought old people eventually understood everything. We don't. There was a friendship I destroyed through pride. A professional decision that hurt someone. Years when I wasn't as available to my children as I wish I had been. An argument with my sister that lasted far too long. HOS could help me understand those things. It couldn't undo them. Sometimes resilience meant repairing. Sometimes apologizing. Sometimes changing. And sometimes it meant accepting that something could no longer be repaired and refusing to let that failure determine everything that came after it. That was harder.
One of my grandchildren asked me a question
She was 19. She had grown up surrounded by intelligence vastly more capable than anything available when I was her age. She asked: “Grandma, how do I know if I'm making the right choice?” I laughed. I told her: “You usually don't.” She looked disappointed. So I explained. Gather what you can. Understand what matters. Listen to people who know things you don't. Understand who else will be affected. Don't confuse what an intelligent system predicts with what you value. Choose. Pay attention to what happens. Take responsibility. Learn. And choose again. She said: “That sounds exhausting.” “It can be.” Then she asked: “Was it worth it?” That question was easier. “Yes.”
The last choices
My world is small now. But there are still choices. What do I eat? Who do I want to see? Do I want another medical procedure? Do I want to talk today or be quiet? What do I want my grandchildren to know? What possessions should I give away? What should happen to the information accumulated about my life? What am I ready to let go? These aren't small choices because there are fewer of them. Some feel enormous. For most of my life, HOS helped me look forward. Now it often helps me look back. But its purpose hasn't really changed. It never existed to create my life for me.
It existed to help me remain as capable as possible of creating it myself.
Looking back
I lived through perhaps the greatest expansion of machine intelligence in human history. Machines became capable of things we once believed required a human mind. That forced us to ask what humans were for. I think that was the wrong question. A human life doesn't require a productivity justification. We aren't valuable because there remain tasks machines cannot perform. My life mattered because it was my life to live. I loved. I chose. I failed. I learned. I hurt people. I helped some. I changed my mind. I raised children who became people I could never have designed. I lost people I thought I couldn't live without and somehow kept living.
I experienced things no prediction could make meaningful for me in advance. Technology gave me capabilities my younger self could never have imagined. HOS helped me use some of those capabilities without confusing them with the purpose of my life. Soon there will be very few choices left. Eventually none. That doesn't diminish the ones I had. If I could speak to the young woman I once was, I wouldn't tell her how to live. After everything I've learned, that would seem strangely disrespectful. I would tell her only this: This life belongs to you. Pay attention. Keep learning. Respect that other lives belong to other people.
When things go wrong, learn what you can and begin again when you can. And don't give away the responsibility—or the privilege—of choosing. I didn't create the life I imagined when I was young. Something better happened. I kept creating it as I went. It was imperfect. It was surprising. It was mine. And I am glad I lived it.
This final story imagines HOS after decades of development. It can understand speech when others struggle to, draw upon a lifetime of information the individual has chosen to preserve, connect health, relationships, history and present circumstances, and help her remember the person she has been. Yet its greatest maturity is demonstrated by what it refuses to assume. Knowing her history does not give HOS ownership of it. Past preferences can inform present choices without imprisoning her in decisions made by a younger self. Predictive systems may conclude that moving is safer, children may lovingly want to persuade her, and medicine may offer additional treatment, but HOS continues to protect the distinction between understanding a person and deciding for that person. Even the accumulated record of a lifetime remains subject to questions of permission, privacy and control. By the end, HOS possesses vastly more intelligence than the early system imagined in the first story, while the individual possesses fewer physical capabilities and fewer remaining choices. Yet the constitutional relationship has not changed. HOS exists to place increasing knowledge and capability in service of the human life entrusted to it—not to become the author, purpose or judge of that life. At the end of life, its success is revealed not by how dependent she has become upon HOS, but by how fully she can still say: This was my life. I chose, learned, changed and chose again.