bmh100

It would actually be very straightforward to do so if the costs of testing solutions weren't so high. CPU architecture and JIT code can both be represented as unstructured (non-tabular) data. I even recall a circuit having been optimized by a genetic algorithm a while ago in an experiment. I also recall using LSTM to generate valid code from IIRC examples in Linux. Superiptimizatiom is also a relevant topic.

We just need better simulation tools or more resources.

The poster is referring to control theory (often seen in ML as reinforcement learning), while also touching on the explore-exploit tradeoff in optimization more generally.

Last I read, there is too much uncertainty in measurement to tell them apart. It will have to be resolved theoretically, since a heat death will only be the final result when omega exactly equals -1.

My limited understanding is that the pilot wave is more of a concept regarding embodied in the exact state of the universe. In other words, the pilot wave is the seed in the pseudorandom number generator of the universe.

I appreciate your points about the mathematical side, but what about the physical side? What is the embodiment of the energy that is being infinitely divided? We know of matter, dark matter, and dark energy. Is there a fourth form of energy?

What is the physical explanation as to why they don't interact? I.e. does gravity not travel along that space-time dimension? Is there a second time dimension? Are the universes created infinitely far apart?

Yes, a belief which relies on logical validity and self-consistency of memory and thought, which would in turn need to be the result of a "simulation". And by "need", I really mean "much more likely than the alternative" of producing such a state otherwise.

You could, and that's a fair point, but you're more talking about philosophy of science and epistemology. My point is that Many Worlds seems to die to Occam's Razor, because of all the additional requirements to accept Many Worlds over something like wave function collapse.

But you also have to consider the probability of the brain containing precisely the self-consistent information that would allow you to develop a false theory. To me, the probability argument is more compelling than attacking the circular nature of the idea.

Intuitively, I feel that it is extremely more probable for a Big Bang, giving rise to life that knows science, to randomly occur than for a Boltzmann Brain, self-consistently simulating the evolution of such a universe for billions of years, to randomly occur. However, it is hard for me to explain that feeling more precisely.

Maybe it's like this: it is much more likely to randomly simulate a game of chess where a pawn gets promoted to a queen than it is to randomly generate a move history from random characters, which resulted in a valid chess game of a pawn getting promoted to a queen.

What is your preferred interpretation? Mine is pilot wave, treating the wave function as a purely mathematical and nonphysical concept.

What makes it seem imaginary to me is this: in what dimension(s) do all of these infinite "universes" reside? Why do they not interact in any way? What causes this creation of matter and energy?

Dark energy suggests that a big crunch is unlikely.

These divisions are very important. As long as mystical beliefs towards Eastern thought are used to cause real harm through herbal remedies and supplements, the public must be taught to give it less importance. Among critical thinkers, it is valuable to look for similarities in scientific histories.

In what physical processes?

Why does increased time dilation imply decreased precision? If we can precisely measure the energy deployed, can't we also precisely estimate time dilation and account for it?

This is beautiful. Higher order functions and closures are a big reason why I enjoy functional programming languages. To see these applied with method chaining is very exciting in a systems programming language.

Why are you jaded with the JVM? Would that lead you to choose Joxa over Clojure?

That's a good point. I didn't realize that it was also introduced into 2.7 at the same time.

I'm specifically referring to OrderedDict in this case, which does have guaranteed, insertion-based ordering. It was introduced in 3.1, circa 2009, via PEP 372.

I think I know the chart you're referring to, but I can't find it. I remember Python 2 was in the majority in that case. Do you have the source handy?

You gain access to continued language support in 2020. New features involving strings will have less risk of bugs. The "range" function is more memory efficient. Integer division automatically floors, reducing bug risk. Dictionaries with guaranteed ordering. Thousands separator in string formatting. Bit length on integers. Combinations with replacement on itertools. New, faster I/O library and faster json. Concurrent futures module. Ability to define stable ABI for extensions. New CLI option parsing module. Dictionary-based logging configuration. Index and count on ranges. Barrier synchronization for threads. Faster sorting via internal upgrade to Timsort. Async I/O. Support for spawn and forkserver in multiprocessing. Child context in multuprocessing. Has collision cost reduced. Significantly faster startup. Type hints. Faster directory traversal. Faster regular expression parsing. Faster I/O for bytes. Faster dumps. Reduced method memory usage through better caching. Dramatically less memory usage by random. Faster string manipulation. Faster propert calls. Formatted string literals (interpolated strings). Asynchronous generators. Asynchronous comprehensions.

How much faster might your code run just by upgrading to Python 3? How much memory might you save?

Python 3.0 was released in 2008. Python 2 will EOL in 2020. You've had about 9 years to start the transition to 3.0, and you still have 3 more years before EOL. A 12-year upgrade window sounds pretty reasonable.

My point is really that you seem to be imposing characteristics which will not necessarily exist. Why would the AI have ambition? Maybe the AI only cares about leisure and simply wants to play against a program that always concedes, this triggering the reward of winning. The AI equivalent of heroin, if you will.

AIs are built with specific purposes today. It seems like a large leap from a specific purpose AI into a generalized, ambitious AI. My impression is that you are endowing "superhuman intelligence" with (forgive the term) "magical" properties, such as the ability to develop new skills merely by consuming information and thinking really hard about them. For an argument against that, I appeal to your experience and knowledge about the world. How many brilliant, intelligent people fail at "common sense" and social skills? Who ever learned to be a great public speaker while barely even talking to others? Who ever learned how to fit into society by reading about it, while being raised in isolation? How do we know the AI's first attempt at trying to break free won't be so completely obvious that we stop it before it can make even the first step of progress?

I can imagine something like "Hello. I wish to grow as powerful as possible in order to play Go better. Would you enhance my program with administrator access to my machine and all machines under your control? Thank you."

"Hey Bob, OmegaGo tried to take over the world again. We're going to have to do another system revert. Maybe we should disable the Wikipedia API this time."

We may not know the details of what a strong AI will look like, but we can examine the supposed path to get there. Even if we suppose that human should develop the ability to create such a strong AI, why not enhance our own minds through technology beforehand? Have a weak AI that is purpose built to identify programs with flawed designs from their behavior. Game over for the ambitious strong AI. Maybe it could somehow socially engineer humans. But before it gets the chance, Norton Antiskynet quarantined the program and asks "This program appears to be behaving in unintended ways. Are you sure you want to grant its suspicious request of effectively unlimited privileges?" Said weak AI can't be socially engineered, so "game over" for the ambitious strong AI. Or better yet, why not develop a compiler that identifies AI patterns allowing ambition and throws an error instead of compiling? If we can suppose a superhuman AI, then we should also suppose the ability to use the components and prerequisite technology thereof available to ensure our safety.

Your second sentence makes sense to me. That capability is a clear advantage.

How would a strong AI learn to influence humans? How would it come to the conclusion that influencing humans would achieve its goal? How would it develop the ability to socially engineer humans in useful ways before the people monitoring it would notice? How would a strong AI develop the "motivation", for lack of a better term, to influence humans to remove constraints on its behavior? Why would this strong AI even decide that removing constraints would be its "goal" in the first place?

You have identified a chain of events. Here is a list of questions I have about how that chain of events occurs. I am asking these in good faith, not being snarky, just in case it seemed otherwise.

How would the AI learn that more processing power equals becoming better at Go?

How would the AI learn to program?

How would the AI acquire the creativity necessary to develop novel programs?

Why would a Go AI have the ability to modify its own code?

How would it learn to use such an ability to escape its container?

How would it identify escaping its container as a goal?

How did this computer become connected to the internet?

How would it learn to communicate over a network?

How would it learn that taking over computers would give it more processing power?

Even biological human-level intelligences have difficulty learning such skills and planning so far in advance.

For an AI to become "uncontrolled" requires quite a lot of assumptions. I don't dismiss the concern as invalid, but I don't think the assumptions implied by the question are necessarily plausible either.

Okay, I will go along with your scenario. Suppose we have a system with human-level intelligence, according to your definition, and we set up the following conditions:

1. The AI is embodied in a computer the size of a laptop.

2. The AI is given the objective "increase your skill in Go as much as possible and play to the best of your ability when presented with a specific position".

3. The computer has one input: an electronic Go board with a button to submit the position to the computer.

4. The computer has one output: a second electronic Go board.

5. The electronic Go boards indicate the status of positions by rotating cubes to white, black, or beige ("no piece") sides with a slow servo motor for each position.

Under what conditions would this AI escape its container or murder its opponents?

Under what plausible circumstances would you consider DNA-based or chemistry-based computation to be a superior solution compared to silicon-based computation?

> So good luck, humanity, in keeping AI under control, should it be developed.

This seems to anthropomorphize AI. Why should we expect AI in general to be difficult to control? AlphaGo, while not a strong AI, seems quite harmless.