3
$\begingroup$

If we make an AI that observe our brain signals for the whole life, won't it be somehow our replicant. That computer would just think like ME. For example that computer should know what I would have decided in a certain scenario even after my death (as we commonly ask 'what would have xyz(a dead person) said in this scenario?').

It learned how I felt in different scenarios of my life for 60 years, that would add some sort of feelings in that machine as well. For example it will be happy on a new born baby and sad on someone's death.

I know if hacked that would be disaster but is it possible?

Note: I am not saying that machine will boost our brain or something (that's another discussion) but just that our brain can live after us and continue learning just like we would have done if we were immortal.

Improve this question
$\endgroup$
8
  • 1
    $\begingroup$ Can you please clarify your question? What do you mean by "replicant?" Are you asking if a computer can simulate/emulate the human experience? If it's possible to "hack" such a program? Further, what do you consider to be a "disaster" in this scenario were the program "hacked?" $\endgroup$
    – Frostfyre
    Aug 13, 2020 at 16:10
  • 1
    $\begingroup$ A many years ago I used to copy books on a copying machine. Despite that machine having observed, and indeed copied, every single word of those books, it showed absolutely no sign of becoming able to write new books itself. Or take this computer on which I am typing: it has observed my typing for many years; and yet it doesn't not seem capable at all to take over the task and type itself. There is a wide chasm between recording physical signals and making sense of them; there is no reason to believe that recording physical signals would automatically lead to understanding the underlying mind. $\endgroup$
    – AlexP
    Aug 13, 2020 at 16:56
  • 1
    $\begingroup$ "It will be happy on a new born baby": why would it? Feelings are in a large part chemically induced. For a taste of the limits of current so-called AI system, read (and understand) Sean Gallagher's article "Twenty minutes into the future with OpenAI’s Deep Fake Text AI" on Ars Technica. $\endgroup$
    – AlexP
    Aug 13, 2020 at 17:08
  • 1
    $\begingroup$ This is the premise of the story “Learning To Be Me” by Greg Egan, and I suggest finding and reading it. $\endgroup$
    – Mike Scott
    Aug 13, 2020 at 17:58
  • $\begingroup$ @MikeScott I second this suggestion (reading "Learning to be Me"). That story was exactly what I thought about when I read the OP's post. And it was one of the scariest stories I've ever read. $\endgroup$
    – NomadMaker
    Aug 13, 2020 at 21:33

2 Answers 2

Reset to default
1
$\begingroup$

PHOTOCOPIES:

You could certainly install a computer network in everyone's brain, recording everything you see in life and analyzing every reaction to every stimulus, then have it extrapolate your predicted actions and compare with what you actually do. Eventually, it could actively and accurately predict all your actions based on how you respond. You would have a machine that could mimic you almost perfectly.

There would be no need for it to have a single emotional response to do so. You always smile when you see a baby, unless you have a headache. You only curse at people who are swearing. A machine could do all this perfectly in the same way a photocopy is a readable replica of an original document. It is static and unchanging. In any predictable situation, your own spouse couldn't tell the difference.

But once entirely novel conditions came up, it would be entirely AI guesswork. The AI would derive replicant action based on how it anticipated you would respond, and as time went on, that anticipation would become it's own thing. Again, no emotion would be needed.If an AI had emotions, it would have it's own responses based on how a machine would respond. Maybe that would be okay. "You" would be a replicant machine, after all, unless the same process mapped out replacement neurons in a manufactured brain to behave like you. This process would be like making a copy of a copy. Opportunities for imperfections would arise. An individual would be born, but it would be less and less like you.

So if you want a truly immortal person, replace the human brain with your AI computer at birth and the copy would BE the original, in a digital format that supported replication. The mind would be vulnerable to change as the substrate of the data changed, but it would be the closest you could get.

Perhaps if you have a simulation that could replicate all the chemical interactions perfectly in a brain, you could copy all the transient energy states in the brain and perfectly duplicate the person in a simulated environment using VAST amounts of information. Even then, you would need to alter the simulation to prevent aging, and then it's not really you anymore.

Improve this answer
$\endgroup$
1
  • $\begingroup$ The Kolmogorov complexity of a human is simply too large for this approach to work. You'd get some bad p-zombie that wouldn't even pass the Turing test. $\endgroup$
    – John O
    Aug 13, 2020 at 18:15
1
$\begingroup$

Only if Tabula rasa theory is correct - and it's pretty much assumed by now that it's not. Also, the AI should be able to pass Turing test with flying colors.

A smart AI can be good at simulating a person, it even should be able to fool that person's friends. But this would never be a perfect replica, just an approximation.

Improve this answer
$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .