Daniel Miessler wrote recently about backing up self to a machine. For the most accurate recording he suggested we could do/add the following:
– Write Extraordinarily Deep Descriptions of You
– It’ll Import from Everything You’ve Done Online
– Journals, Texts, and Other Private Data
– Extensive Interviews and Scenario Exercises
– Interviews With Loved Ones, Friends, Coworkers, and Associates
– Your full Genome
If you performed all these steps/methods, is what results a human-type consciousness? (Forgive me, Daniel.)
Sorry if this is an old argument, but it’s important. Whatever resulted would be a binary-stored collection of attributes. ChatGPT has all the answers that dance around the question of consciousness. We chat with it and actually think that it’s the same as a human. (Miessler did not say that either.)
A piece in Wired by Meghan O’Gieblyn about ChatGPT suggests that humans conceal the “shadow” underneath through a mask of socially acceptable qualities. To sustain the mask we have also given rules to LLMs so they self-censor; their security walls can be bypassed. The chatbots do not have “intrinsic agency or desires.” The chatbot is only “trained to predict and reflect the preferences of the user. They also lack embodied experience in the world.”
O’Gieblyn says AI can be closer to a conscious being if we input into it a subliminal realm through our cultural repositories and the archetypes we know from “remixes of tropes drawn from human culture, amalgams of our dreams and nightmares.” The machine convinces, so we impute to it our humanity by giving our creation “fears and longings that we are often incapable of acknowledging to ourselves.” All these inputs are not sufficient.
O’Gieblyn is correct, “we are more than a collection of irrational instincts–or statistical patterns in vector space” that “allow us some small measure of agency over the mysteries that lie beneath.” The unpredictable, the subconscioua–we are trying to give it humanness. Will the mystery of consciousness be solved in our lifetime? It seems we want to impart our souls into the machines, but don’t know what that is. We can give it fears and longings and feed it data en masse. When will its response be anything new? ChatGPT is already an unoriginal writer.
The Stereo MCs’ song Connected speaks about the gaping hole in reality (that was the 90s, so I’m dating myself). We can gather all the records, do the interviews, write journals and make podcasts–but will the human LLM convince the survivor? We have to rely on those things to bring remembrance of absent others, someday to convince us. But our desires will be for more and more reality for these personas. New. Unpredictable. Though the unspoken will be more and more spoken, the realization of humanness fleets away. After days (months, years) the story will fade.
Faith, not just faith in a religious sense, but in mankind, answers and does those things that only a real person can do. It imparts beauty, not just repetition.
Someday I may be wrong. Then this article will be written over and over.