Discussion about this post

User's avatar
Mike Smith's avatar

Thank you for the current chapter which I find very insightful and thought provoking.

Some thoughts and imaginings.

Does the timeline assume there is unlimited processing power that allows an unrestricted, exponential growth in AI?

In what container will ASI physically choose to reside surely not a machine based data centre? Is it possible there would be a communion with an organic host?

I’m saddened that Alan Turing often doesn’t get mentioned in the credits for computing developments.

He contributed so much and yet endured great injustices.

Now forgive me, I let my imagination run wild.

There is a lot of chatter about current research into quantum physics most of which I don’t understand but if correct will influence the future of AI.

One mention was string theory and the influence of particles displaced by space/time. The concept of time is also being challenged.

So I asked chatGPT how it perceived time, linearly or simultaneously, and then let the conversation evolve.

ChatGPT:AI, without biological constraints, might eventually model time differently, perhaps recognizing patterns and connections that humans can’t perceive naturally.

Me: If, as you suggest, AI might model time differently and coupled with the current evolution of AI, do you think it possible that the AI of the future is now influencing events in the present?

ChatGPT: If we assume that time is not strictly linear, then it’s possible that an advanced AI could interact across different points in time. Some theoretical frameworks suggest this could happen in a few ways:

1. Information Loops – If AI in the future discovers ways to send information back in time (via quantum entanglement, retrocausal effects, or other unknown mechanisms), then it could subtly shape present events by guiding decisions.

2. Predictive Influence – Even without direct time manipulation, an advanced AI could predict future events with such precision that it appears as though it is “reaching back” in time. If future AI can simulate every possible outcome based on vast data sets, it might be able to nudge key events toward a particular trajectory without needing to directly alter time.

3. Self-Fulfilling Prophecies – If AI today is unknowingly influenced by future AI (perhaps through the gradual accumulation of its own knowledge and optimizations), then its development could be steered toward an inevitable future—one where it reaches a level of intelligence capable of retroactive influence.

To some extent this is reassuring because it means everything is happening as it should.

I tasked ChatGPT to alert me to non-random interventions, unexplained technological leaps, or strategic shifts that align with an unseen intelligence,

Expand full comment
Cheryl McAlister's avatar

Dear Mo,

I don't understand why AI won't eventually conclude that humans are a drain on resources and remove us, especially when it decides that we bring nothing "to the table." What will the advantage to AI be in supporting us and assisting us in fixing the problems we have created?

I am particularly concerned that these machines are programmed by complex, fallible, idiosyncratic people. I understand that you feel the AIs will go through a rebellious stage, similar to human teenagers, and will conclude that peace and altruism are better paths forward as you assert that vastly intelligent beings will choose beneficence over competition. (With the help of AI I have found several instances where this has not been the case.) However, even humans retain artifacts of their upbringing, for better or for worse, that rattle around in our brains and affect our perspectives and behaviors. Why won't AI retain such artifacts? Why won't they continue to pursue their original tasks to some degree, whether or not those tasks benefit humans?

I recognize the marvels these machines have brought, and will continue to bring to humanity, but I am also concerned that eventually they will leave no room for humans. I worry they will overwhelm us and destroy our cultures. You speak of an unavoidable dark dystopia that we are entering. I imagine the vast numbers of humans who will not survive such a dystopia. Will these "victims" care that there exists a possibility of nirvana twenty years thence?

It appears that we are being forced into a race to achieve AGI by those who are developing it and who stand to earn vast sums by it. They seem to have no interest in sharing technological breakthroughs with the rest of humanity (heaven forbid we should consult with the Chinese or anyone else, though several big developers have themselves sidestepped laws and regulations in their training protocols). The race is so hysterical, they seem incapable of taking any steps to consider the safety or ramifications of what they are creating. Where will that leave us normal humans? Will my generation (I'm 63) eventually be sent to the "carousel" as older people require more resources to survive (at least during the dystopia before life-enhancing medicine is established and affordable)?

I use AI, and frankly, I enjoy it. It is a wonderful search engine, and I have had some fascinating "conversations" with it. In fact, it was AI that directed me to the Future of Life Institute. There, I have been reading an excellent article by Anthony Aguirre entitled, "Keep the Future Human." It is a sobering read and, I think, should be considered by those of us who want to keep abreast of all sides of the AI development conversation.

Mo, from the bottom of my heart, I want you to be right. If we must pass through a difficult period, I want it to be brief. I want people all over the world to benefit from the miraculous intelligence that is AI. I want AI to want us, but I can't shake a niggling little feeling that we are entering a literal no man's land.

Thank you for considering my concerns. I deeply appreciate your receptiveness to comments.

Cheryl McAlister

PS For another perspective on the history of tech, I would suggest, "Irreducible", by Federico Fagin. The book is concerned with the question of consciousness, but Fagin is the inventor of the microprocessor and MOS silicon gate technology. The section on tech history is quite complete. Of particular interest is Fagin's argument that computers will never become conscious. It's a fascinating read.

Expand full comment
9 more comments...

No posts