Level1 News October 31 2018: Tesla Gets High

Twitter isn’t a public utility. Its a corporate entity and should be allowed to make its own choices.

I mostly fail to understand why you " as the people " continue to use such sites that continue to openly abuse such things.

It is easy… stop using such services and watch them die.

But you will not ? Like a bunch of junkies seeking your next idiotic fix.

1 Like

IMO I think its time for a Digital Bill of Rights

I mean lets face it the EU is doiing a terrible job with the internet. ISPs in the US are pretty much screwing you guys over. If there was a set of inalienable rights for the use of the internet then a lot of these problems wouldnt exist.

I’d like to weight in/yell into the void on the DARPA “contextual AI” thing. I don’t believe what they’re talking about with that term is Artificial General Intelligence (AGI), though I’ve no doubt they would love to have that in a few years.

What they’re talking about is something I’m quite interested in: combining the symbolic approach to AI - the old, original approach of John McCarthy origin, using structured formal languages to model and reason about the world; and machine learning as it exists and functions today - learning from lots and lots of data, and accurately predicting or classifying things in the world.

At present, ML tries to guess what things are, and with a certain confidence, even if it hasn’t seen anything like it before, because that’s what it’s trained to do. It’s trained to generalise, over the training set, to things it hasn’t seen before. The problem then comes when it sees something unrelated to everything it’s seen and it goes ahead and classifies it anyway, perhaps even with a seemingly large amount of confidence.

Instead, if we can build rich-enough models of the world, and still train an AI but within that world, it might be able to reason that something has not been encountered before and is entirely novel to its environment.