News for hackers. Stuff that matters.

KellhusSmellhus

What I read from the article was that high-level models are more than just compression - a device for a La-Place-demon to cut on computation time.

But I also have a strong suspicion that this is not what they meant; maybe someone can clarify? In the other case (my interpretation being in line with the intent of the authors): How do they justify this?

Ask HN: What does HN think of a intelligence agency DAO?

My thinking, which is not new, is this: If we are being spied on, at least let us spy back. Also, most people would be more motivated to take care of their privacy if not only big brother but i.e. their real big brother can read their browsing; this is interesting for everybody since a controlled society is a worse one, best example: You don't have WhatsApp? You don't get invited by new groups...

( Read More... | no comments | on HN )
There is already brain-changing tech, drugs.
But money spent on space interested experts could be spent on climate interested experts.
I wonder whether those experts and resources used on space could do feasible climate engineering...

The only environmental existential risk (apart from the human extended phenotype of course :) ) is climate-based, no?

As long as the resources are not orthagonal the problems are not orthagonal. Meaning the money and effort shooting rockets could be put into fixing the environment, the "defense" problem, and all kinds of suffering and death.

One could argue that space could give us a shitload of resources to solve those other issues.

On a related note I would like to plug the meme that "humanity" as a whole is worth shit; what matters is the qualia of actual humans and then other possibly conscious beings. I don't care whether there are humans somewhere in space if we down here die from nuclear holocaust, global warming, aging or hunger. I am not "humanity" and neither is anyone else. We are all individuals striving to survive and not suffer, and sure we can cooperate in that endeavour and create superorganisms to aid us - but the emphasis is on "aid", which is not "enslave". I shit on any superorganism building meme that does not put actual humans or consciousnesses first.

EDIT: typos

Re:Concrete AI tasks for forecasting (2014) on Friday January 06, 2017 @04:20AM (#13334916)
AGI is a pretty hard to define term, but I never saw it defined as "self-conscious". Most definitions are along the lines of "do what a human does", i.e. walk into a random house and make coffee or get a college degree, or, more mathematically, "Average a good score over all possible problems, weighted by their likelyhood and the compactness of your code"
Re:Tool AIs want to be agent AIs on Wednesday December 21, 2016 @09:00PM (#13234205)
Is this what they call anatta?

How does one get into AGI/ASI research as fresh graduate?

I am almost done with my master's in compsci and facing now the decision how to contribute to the only thing I consider worthwhile - ASI. Here are the paths I am aware of and their dangers/difficulties from my pov (I should clarify that I think the path to ASI leads over AGI, so I'll use those terms interchangeably):- Machine Learning Research: Relatively far from AGI, and everybody and their...

( Read More... | no comments | 2 points | on HN )