OpenAI: Gathered Articles from the last few hours (or a Mini-Mega-Thread)
We had a thread about OpenAI Staff Threaten to Quit Unless Board Resigns, but I thought I might as well add it again. Especially because of this part:
Remarkably, the letter’s signees include Ilya Sutskever, the company’s chief scientist and a member of its board, who has been blamed for coordinating the boardroom coup against Altman in the first place.
Okay then. I think we are in a simulation, someone quick saved, and is now experimenting what the outcomes of random decisions are.
A minor piece of information was that OpenAI Approached Anthropic About Merger, and The Atlkantic has a slightly longer look and speculation what’s going on Inside the Chaos at OpenAI.
With Ilya’s recent turn around, there’s apparently also the option of Altman coming back Sam Altman is still trying to return as OpenAI CEO, something even MS would apparently be okay with, at least publically.
Business Analysis blog Stratechery posted some analysis on OpenAI’s Misalignment and Microsoft’s Gain.
Loving it, this is like SubredditDrama, but without having any actual chance of affecting me (I don’t believe in AGI coming out of LLMs), and on a global scale.
Rather unlikely, as the multiverse theory is a...theory.
I reason it like this.
Do you believe it is physically possible for an extremely advanced enough civilization to build a simulation where conscious simulated beings live unaware that they are simulated. Yes|No
If the answer is no, well then you can stop reading because this wont interest you further.
If the answer is yes then you may agree that over all the time in the universe bang to heat death there will be at least once such civilization that gets to this level.
If they get to this level running such simulation it is not unreasonable to assume they wont stop to only run a single one. Part of the usefulness in Simulations is that you can run many next to each other and if you have the knowledge and means already then why not.
So if you answered yes in the first question you are now at a not unreasonable hypothetical of at some point in all of time a real world society is running many simulated world.
In many ways one can argue that many simulated worlds is already a multiverse.
But lets continue with a conservative definition and say simulated world is not a separate universe,
there can only be one true real world. Which at some point may run simulations that can hold conscious lives.
You are a conscious entity, you believe to be born in this world but you never knew any better or else.
There are near infinite locations and times in the real world where you could have be born/started to exist
But a single simulation already doubles all of time and space, there are infinitely more locations and times your consciousness could manifestation in one of the many simulated worlds.
So getting to this point. What are the odds your conscious manifested inside a human embryo during one of the most interesting scientific times of human history and it also being the one real non simulated world?
Probably pretty good odds were not.
The laws of physics we observe have far to much unnecessary complexity at the small scale to be worthy simulating unless it mirrors the simulators reality, in which case it’s unlikely that you could simulate earth on anything less than an truly absurd collection of matrioshka brains, and even then it would be to expensive to do often.
Moreover, what would actually be the point of an ancestor simulation in the first place? Things are going to diverge so rapidly that the only things you can learn are very general statistics, which you could make good estamates of already and with far less computing power. These are also statistics that self evidently do not matter becuse if they did occur often enough to matter than the simulators would already know just by looking around their own universe.
Basically the only things i’ve heard that don’t require an intelligent civilization that could reach technology far in advance of our own but is comprised completely of people so psychopathic they’d create billions of children just to kill them off for mad science are things like your parents wanting to raise thier children in a simulacrum to the distant past or wiping your own memory for some roleplay or similar. In that case, where you are only simulateing a few rea people and a bunch of NPCs, why would you bother designing NPCs to being up the simulation in the first place?
Foreshadowing and getting you used to the idea maybe, but there are a lot better ways to do that, most of which involve an actual conversation.
Quick note that this is just a hypothetical exercise i like doing. This world is as real as it will get for the duration of my life.
Interesting point that a society so advanced may not have much to gain from ancestor simulations. I havent heard that one before so thats an interesting for me to get into some night. I would think that if they may at some point fully have explored all their is in the present they may look to the past and alternative universes out of boredom.
Nonetheless I believe there is a very good reason to run the if you flip the maliciousness from you story upside down into a loving gift of life.
Why is it so unnecessary complex?
Because only the best will do for life, no one wants to live in a low powered simulation. If a simulated world is not as complex and deep as a real universe can we expect the experience and lives of its inhabitants be? We are studying those unnecessary complex phenomena right now, they may cause the next technological breakthrough.
If they where not there and while searching we find one of the tubes carrying ones and zeros we might have freaked out before maturing.
What could they still learn?
I would usually assume they may simply try to learn the conditions that lead to the big bang by simulating all kins of possibilities till you have one which is a close enough copy of reality.
But in this hypothetical we assume they are passed needing to learn anything. So they don’t instead it may simply be to teach “life” and “connections” are things that “exist” and can be “experienced”. The simulation is a loving nest designed by our techno superior parents who wish to nurse new complex life in the safety of their hardware.
It may also be that we have a purpose in the real world and that we are AI in training. The best way to solve the alignment problem is to give each AI a full experience of life with goods and bads.
Why would they do this over biological offspring?
Because (pure opinion) time is the real final frontier. Eventual cosmic heath death appears inevitable. Living near forever without aging may be possible but creating more time in the universe may not. Simulations allow life to experience much more time then outside of them. So not only are we reasonably safe in the simulation we also have all the time we need.
Of course i have conveniently ignored all the suffering in our world in this loving gift but in this context that depends on perspective. Was the human world designed as part of the simulation or was it organically generated as we evolve? Are we a single conscious in a human world? A human hive mind consciousness on a planet or are we all part of a bigger consciousness encompassing our universe. Maybe those outside our world may know, but I definitely do not.
Would it be also evidence for the simulation theory that there is a maximum speed of information (light speed)?
I am as much a astrophysicist as the average lemming but is it ever proven that there is a maximum speed of information?
It may depend on how speed is defined. Maybe a maximum movement speed within space time has such limit but the fastest way to go from a to b can still be instantaneous through quantum mechanics.
Big think has an article on quantumparticles, the way I understand it they are the smallest bit of information that could potentially be both a proton or electron. I found it sounds a bit like a “bit” having the potential to be either false or true.
I think if any sciences proof the simulated world theory its quantum mechanics excluding all other possibilities or ASI figuring it out and telling us.
Sorry man, I misread, my answer is BS.
So is optics. So is Newtonian gravity. So is relativity. So is quantum mechanics. So is... Everything. This is a gotcha on the level of "it's made of chemicals".
In this context, "theory" doesn't mean "unsupported idea". It's a framework for understanding observable evidence, that is often well supported by evidence.
Fortunately, "multiverse theory" isn't a theory...
Sorry guys, I misread the post before. Please ignore me.
Well at cosmological scale we do not have anything better....