All of Akira Pyinya's Comments + Replies

Thank you for your introduction of Richard Jeffery's theory! I just read some article about his system and I think it's great. I think his utility theory built upon proposition is just what I want to describe. However, his theory still starts from given preferences without showing how we can get these preferences (although these preferences should satisfy certain conditions), and my article argues that these preferences cannot be estimated using the Monte Carlo method. 

Actually,  ACI is an approach that can assign utility (preferences) to every p... (read more)

Thank you for your reply!

"The self in 10 minutes" is a good example of revealing the difference between ACI and the traditional rational intelligence model. In the rational model, the input information is send to atom-like agent, where decisions are made based on the input.

But ACI believes that's not how real-world agents work.  An agent is a complex system made up with many different parts and levels: the heart receives mechanical, chemical, and electronic information from its past self and continue beating, but with different heart rates because of ... (read more)

Yes, of course some species extinct, but that's why organisms today do not have their genes (containing information about how to live). On the contrary, every ancestor of a living organism did not die before they reproduce. 

I think I have already responded to that part. Who is the “caretaker that will decide what, when and how to teach the ACI”? The answer is natural selection or artificial selection, which work like filters. AIXI’s “constructive, normative aspect of intelligence is ‘assumed away’ to the external entity that assigns rewards to different outcomes”, while ACI’s constructive, normative aspect of intelligence is also assumed away to the environment that have determined which behavior was OK and which behavior would get a possible ancestor out of the gene pool.&nb... (read more)

Thank you for your comment. I have spent some time reading the book Active Inference. I think active inference is a great theory, but focuses on some aspects of intelligence different from what ACI does.

ACI learns to behave the same way as examples, so it can also learn ethics from examples. For example, if behaviors like “getting into a very cold environment” is excluded from all the examples, either by natural selection or artificial selection, an ACI agent can learn ethics like “always getting away from cold”, and use it in the future. If you want to ac... (read more)

In the post I was just trying to describe the internal unpredictability in a deterministic universe, so I think I have already made a distinction between predictability and determinism. The main disagreement between us is that which one is more related to free will. Thank you for pointing out this, I will focus on this topic in the next post.

I want to define "degree of free will" like:  for a given observer B, what is the lower limit of event A's unpredictability. This observer does not have to be human, it can be an intelligence agent with infinite computing ability. It just does not have enough information to make prediction. The degree of free will could be very low for an event very close to the observer, but unable to ignore when controlling a mars rover. I don't know if anybody has ever described this quantity (like the opposite of Markov Blanket?), if you know please tell me.

I like... (read more)

You are right, this picture only works in an infinite universe.

Thank you for your comment. Actually I am trying to build a practical and quantitative model of free will instead of just say free will is or is not an illusion, but I can't find a better way to define free will in a practical way. That's why I introduce an "observer" which can make prediction.

And I agree with you,  claims like "not 100% correctly" are too weak. But possibly we can define some functions like "degree of free will", representing how much one subject could be predicted or controlled. I'm not sure if this definition resembles the common meaning of "free will", but it might be somewhat useful.

Thank you for your comment, but it would be appreciated if you could prove my conclusion is wrong (e.g. either observer B1 or B2 is able to know or predict event C)

Sorry for the misleading, but I also believe that libertarian free will is not an illusion. I hope I can explain that in the next post on this topic.

(Maybe I should add a "(1)" behind the title?)

You are right, I should use "all initial state on a given spatial hypersurface" instead of "all causes", but the conclusion is the same: wherever the hypersurface is, no observer is able to know all the initial state on that hypersurface which can affect event A, except when the observer is in the future of A. 

The second question, I think that "high accuracy" is only the upper limit of a prediction, which is not that easy to reach. In oder to make high accuracy prediction, you need a large amount of resources for observations and calculations. The amo... (read more)