Hacker News Reader

Hacker News

Comment by indrax | Hacker News Reader

Parent Full thread

indrax

I didn't downvote you, but your argument is terribly flawed, and has been answered thousands of times.

* You give no justification for rejecting powerful AI

* Many AI projects use robots in development

* An intelligent agent does not need a robot to get things done.

* Many AI projects supply Internet access to the AI.

* A sufficiently intelligent agent might circumvent protections you think are adequate, or manipulate you into doing so.

View on HN

  > You give no justification for rejecting powerful AI

I wouldn't go that far as to reject it completely, but it will take a really really long time, orders of magnitude longer than Krzweil imagines. And the reason is very simple—we have no idea how mind really works. Our current knowledge is like the knowledge of the early alchemists compared to the current chemistry, only millions times worse.

Speaking from a background in neuroscience, that simply isn't true. We are still very ignorant when it comes to the workings of the brain, but we are far ahead of alchemists, simply because we are approaching the problems empirically, which alchemists did not do. Of course, the brain is many orders of magnitude more complex than chemical processes, so in that sense we have a much longer road ahead of us. We do, however, have the advantage of modern technology to accelerate our progress.

But the most important factor by far as to why I believe you're wrong is that we do not have to understand how the mind works on a deep level to create AI. Our brains do a ton of stuff that an AI doesn't have to. We only need a fairly basic understanding of the brain to create thinking machines.

Seems like it can't really be all that complicated (really!) The brain was invented by evolution (the lowest bidder, essentially) which operates blindly. The thinking mind was invented maybe 1 million years ago, by an evolutionary drunkards walk around the neural-connection-space. There's got to be some relatively simple structure involved, recursive or iterative or just random connections that learn?

The "thinking" mind is one of the simplest parts of the brain. Ironically, what philosophers call the Hard Problem of Consciousness is much easier than the "easy" problem. The mechanisms by which we perceive the world are vastly more complex than the mechanisms by which we process and store our perceptions, and those parts of the brain have been evolving for hundreds of millions of years. The sheer amount of optimization is what makes it so hard to replicate. We can make artificial systems that outperform tiny invertebrates, but we still have a way to go. Fortunately, our rate of progress is rapidly accelerating, so I'm still confident we'll be able to figure it out in the next few decades.

Reply on news.ycombinator.com