I wonder how long it will take before the government decides to give up on trying to use “proper” ways to get this done, and just uses the FCC to create new regulations, requiring providers to have backdoors, etc. Maybe it’ll be in the next Net Neutrality rule set.

> That is a supposition. Where is the data? What suggests a creator?

I never admitted to not reading the article, and I have read the article. Don’t give me any of that bull shit. I have also explained repeatedly why the study, regardless would have still been valid, if the authors actually did mean “creator.”

It plays better than one person, in real time. That’s it. Why? Because computers can do automated tasks faster than people can. So what? That’s not intelligence.

I don’t think so. The discussion is about an AI doing something. But I really don’t consider machine learning like that to be AI. AI used to mean what we now refer to as AGI (artificial general intelligence). Machine learning is just task automation.

You’re right. When you first replied to my comments, I thought you had sufficient reading capacity and intellect to engage in a discussion. Now I know that I am wrong.

My point is the one that I made. They made a slightly faster search algorithm, which searched for a specific set of criterion. It’s somewhat interesting, but in no way impressive.

> You proved it? Proved? haha. You won’t even take 2 minutes write it down for me.

> I don’t understand your clinging to this line of criticism. I don’t believe anyone has ever made the claim that this program, once it is able to beat a human Go champion, will go on to pilot aircraft, or do brain surgery, or even play checkers.

I never lied about my credentials. I am well educated, and have had to read through many papers, in many different fields, in order to get my degrees. No; I am not published in a peer reviewed journal, and never claimed to be.

Probably. They’re very good at learning automated tasks and using massive amounts of existing data and processing power to do something we can do, but much faster.

That’s not what I read. They used an already existing data set, and a tree search. It then relied on Google’s AI architecture to cut out search paths that were likely to be useless. The AI may have learned how to better navigate the tree, by playing repeatedly, but it did not learn how to play go. Obviously, otherwise…

I can only be polite for so long, especially when others really haven’t afforded me the same courtesy.

> I wouldnt say infinite, we tend to forget things when we dont practice them regularly, like language.

> automated model fitting is what athletes are trained to do. Think an NFL running back that has to recognize where to run in the matter of a split second, or a tennis player that must recognize the best place to return the serve or volley of the opponent before the ball even reaches their racquet.

AI used to be what we now call AGI (artificial general intelligence). It’s very different from automated model fitting. True artificial intelligence requires being able to learn new tasks, and incorporating novel information and information classes, on the fly.

You’re not getting it. The system was developed to learn one task and one task only, and that’s all it can do. Some algorithms, powerful computational power, and huge data sets were used to simply find a solution to a very specific problem and that’s all the system can do: play go.

The rules are simple. Playing is not.

I see what you mean by “novella.”

I try.