Hacker News Reader

Comment by wmat | Hacker News Reader

Isn't it also just a matter of time before machines are writing the software?

Programming isn't "writing software". It's "understanding humans and translating what they want into algorithms". I don't think it's possible to automate that without general AI (at which point, writing software is just telling the robot what you want).

If that became true, I'd just switch to basic research. There are an infinite number of open-ended questions to answer in the universe.

If machines become better at basic research than humans, then we have achieved one of the key features of Banks' Culture. That's a feature, not a bug.

Whoa there. I think there are a certain, countable number of fundamental questions to answer about the universe. And you can count them with your fingers and toes. They are the fundamental constants, and they're already known to 3 or 4 decimal places.

Albert A. Michelson:

"The more important fundamental laws and facts of physical science have all been discovered, and these are so firmly established that the possibility of their ever being supplanted in consequence of new discoveries is exceedingly remote."

After that, he lived just enough to see the physics giving birth to modern chemistry and electronics.

Imo the flaw in this reasoning is, altough we may know an aweful lot about chemistry and the physics of atomic and molecular interactions, the problem of predicting the folding of proteins from the sequence of amino acids is exceedingly difficult. This is why we still dont have custom engineered protiens.

This is an example of a highly important and interesting problem that is limited by computational ability, among other things. Knowing some physical constants is important but the macro problem is still hard.

The bottleneck in basic research is not employees - we already have too many of them - and we're already starting to automate lots of the research work,both the manual one(robotics in biology), and the mental one(lots of the data analysis in many scientific disciplines could-be/is automated).

Creating the Culture is some feature. Heck even if it was the result of a bug it would be the best bug ever.

For those of you who haven't read Ian M. Banks "The Culture" series, The Culture is a space-habitating multi-species confederation that has long ago made true general AI, concurred all ageing and illness and now mostly spend their time exploring the universe and trying to improve other civilisations, without too much meddling. Anything at all we can do to become them is a Good Thing.

No. Programming is easily hard enough that any software that can do so alone (as opposed to with a human) will be a human level AI.

At that point we just found The Culture and start to build space-ships.

The incentives for this aren't very good.

Reply on news.ycombinator.com