So, I had an idea about making an AI bot that can be “like me” to a great extent. Now the obvious and the hard way would be to make a system that is able to interpret its environment, make everyday decisions and respond to external stimuli. However, what if we try to take a shortcut and make a system that can mimic different aspects of a person’s personality with high fidelity.
For example let us consider a case where a system is tasked with mimicking heartbeat patterns in response to different stimuli (audio, visual, touch, smell, etc.), i.e. if we monitor how a person’s heartbeat changes and gets affected under different scenarios, a relatively simple model should be able to mimic that pattern (making a reasonable assumption assuming the feature set is sufficient and there exists a correlation between external stimuli and heartbeat of a person).
Now my real question is what if we extend this idea to mimic not only the ECG patterns (heartbeat) but also the EEG patterns (brain wave patterns). A similar approach of recording a time-series snapshot of brainwaves given external stimuli (a much more complex and controlled environment is assumed to reduce the effect of external factors such as temperature, smell, etc). Finally, if the above-mentioned approach seems to be feasible, if we create multiple AI agents each mimicking brain-waves for particular stimuli or maybe even a particular region of the brain can we create a model capable of mimicking the personality traits of a person (maybe something between low and high fidelity)
Yes, I understand even if the stuff mentioned above is true the model is not actually conscious rather just really good at following a pattern, however, even a model that can mimic the traits and actions of a person to a great accuracy should be good enough.
P.S. I apologize in advance if the aforementioned idea seems crap and illogical, I just had to get this idea out of my mind and have a discussion over it.