A possibility of Eternal Life Through Machine Learning

Artificial Intelligence

Recently, I watched the movie Man of Steel (a bit late, I know). The holographic afterlife form of Jor-El, played by Russell Crowe, piqued my interest and made me wonder whether it might be possible to store a human personality like a computer program. Such a program could react and behave according to external conditions as the real person would. After brainstorming with my brother, I found the concept feasible and decided to write an article about it.

For those who haven’t seen the movie, after his death, Jor-El communicates with his son, Superman, in the form of a hologram. Before dying, Jor-El manages to store his consciousness in a device similar to a flash drive.

If I had watched this movie a few months ago, I might not have thought about these ideas. However, I recently completed a Machine Learning (ML) course on Coursera. Machine Learning focuses on constructing mathematical models and training them based on past data (input-to-output relationships) to predict outcomes based on new inputs.

If you haven’t heard of this term and are not from a computer science or mathematics background, let me explain it with a basic example. Consider the following equation:

Output = A * Input + B

Given values of A and B, this equation can provide outputs for different inputs. This concept forms the foundation of a model called Linear Regression, which is used in Machine Learning for predictions. The key is finding the right values of A and B, achieved during the training phase with a set of input and output data. The model aims to find the values of A and B that best fit the equation.

Although this may require a leap of faith, such models, with enough training data and tuning, can predict many outcomes with high precision. Google’s driverless car uses Machine Learning to train a model that learns how to drive by observing a human driver.

If the training data comes only from one person, the model will replicate that person’s driving style. For example, if the model observes your dad driving a car, a ride in a vehicle driven by that model would feel like your dad were driving.

Now, imagine a mathematical model that captures what your dad would say in response to specific conversations.

Dad-Speak = A * Dad-Hear + B

Training this model would involve recording your dad’s voice and the conversations he hears.

Can we go further and create a model that replicates not just the driving style and speech but the entire personality of a human being? Imagine capturing the facial reactions and body language of someone when they are angry, sad, or happy. The output could be similar to Jor-El’s holographic projection, displayed by a computer program based on a Machine Learning model trained to capture the person’s nature’s nuances.

The possibilities are endless! If we could create such a model:

• We could model a scientist like Einstein to solve research problems or continue researching as if he were alive.
• In the event of an accidental death of a country’s president, such a model could govern the country as the president would, temporarily preventing chaos.
• You could converse with loved ones, see them, and receive their opinions or guidance on important matters, even after they’re gone.

The Linear Regression example in this article serves as an illustration, as more sophisticated models like Neural Networks exist. More research is needed to identify the right model and gather sufficient data to train it.

I don’t know how long it will take to create a model capable of accurately capturing so many aspects of a human personality, but I’m fascinated by the idea and the potential for eternal life through Machine Learning. I hope such a model exists one day that I can train to behave like me before I pass away.

PS: By eternal life, I mean living on for others and your loved ones, not living forever.