AI Correlation
Posted: Wed Jan 03, 2024 8:11 pm
The most likely revival scenario is digital rather than physical. This is also mostly likely to happen concurrently with the Great Upload, when all living humans are also uploading over a fairly short time period of less than a few decades. But why would a future society want to pay to upload underfunded brain preservations, let alone masses of very poor living humans? I think there will be a huge incentive in the form of AI Correlation.
Let me explain. After they revive my mind, I will be left with the same level of intelligence and memory recall as I currently have. That’s not great and we could do so much better. An AI could easily extract each historical memory from my mind and tag it with a specific GPS coordinate and date/time. For example, I remember riding a bicycle with a particular group of friends once as a child. I do remember the exact location, but the date and time are not attached to that memory. It’s going to require some correlation to properly date that event. This would be far too tedious for humans but very easy for AI. For this event, it might look at the specific bicycle I was riding and what date range I owned that bicycle. I also know the parking lot was empty, so that eliminates certain dates. The weather eliminates other dates, etc.
But here’s where it gets interesting. I remember at least one other specific individual in that group. If that person uploads and shares a carefully sanitized version of all their memories, then we can begin to have some correlation. That other individual wrecked his bike and probably even has a resulting scar. His AI can bracket the date of the event a lot tighter, and it can probably also add at least one more individual and a bit more detail to that shared event. That additional detail would very likely jog my memory and make it richer. In this manner a shared history can be built, and the more people who help, the richer it will be. What this means is that the millions of random people who I've interacted with over the years are all important to me because they can help rebuild my personal history.
I have incentive to revive others and I think it will be a very strong incentive. I think this will become immediately apparent once my AI has properly organized my memories and hooked them onto dates and times. I will be frustrated by missing details and wide brackets on many of the dates, and I will naturally seek more information. Some of the best correlation will obviously be documents, databases, images, and videos. But other minds will also provide rich correlation. And because of the way that humans are connected, this will quickly expand to include all of humanity. We will want to upload everyone for entirely selfish reasons.
In addition to enriching our personal histories, this will also deeply enhance our general historical knowledge. I enjoy my current ability to use google maps to walk around in nearly any city in the world. Imagine how much richer that could get over time, with the ability to move through space and time. My AI narrator could be telling me stories about what I’m seeing to help me understand the context. It would be kind of like a limitless personalized documentary. With some memory contributions, these reconstructions would be much more interesting.
Let me explain. After they revive my mind, I will be left with the same level of intelligence and memory recall as I currently have. That’s not great and we could do so much better. An AI could easily extract each historical memory from my mind and tag it with a specific GPS coordinate and date/time. For example, I remember riding a bicycle with a particular group of friends once as a child. I do remember the exact location, but the date and time are not attached to that memory. It’s going to require some correlation to properly date that event. This would be far too tedious for humans but very easy for AI. For this event, it might look at the specific bicycle I was riding and what date range I owned that bicycle. I also know the parking lot was empty, so that eliminates certain dates. The weather eliminates other dates, etc.
But here’s where it gets interesting. I remember at least one other specific individual in that group. If that person uploads and shares a carefully sanitized version of all their memories, then we can begin to have some correlation. That other individual wrecked his bike and probably even has a resulting scar. His AI can bracket the date of the event a lot tighter, and it can probably also add at least one more individual and a bit more detail to that shared event. That additional detail would very likely jog my memory and make it richer. In this manner a shared history can be built, and the more people who help, the richer it will be. What this means is that the millions of random people who I've interacted with over the years are all important to me because they can help rebuild my personal history.
I have incentive to revive others and I think it will be a very strong incentive. I think this will become immediately apparent once my AI has properly organized my memories and hooked them onto dates and times. I will be frustrated by missing details and wide brackets on many of the dates, and I will naturally seek more information. Some of the best correlation will obviously be documents, databases, images, and videos. But other minds will also provide rich correlation. And because of the way that humans are connected, this will quickly expand to include all of humanity. We will want to upload everyone for entirely selfish reasons.
In addition to enriching our personal histories, this will also deeply enhance our general historical knowledge. I enjoy my current ability to use google maps to walk around in nearly any city in the world. Imagine how much richer that could get over time, with the ability to move through space and time. My AI narrator could be telling me stories about what I’m seeing to help me understand the context. It would be kind of like a limitless personalized documentary. With some memory contributions, these reconstructions would be much more interesting.