News for hackers. Stuff that matters.
Re:Programming won’t be automated, or it already has been by mtdewcmu on Wednesday March 22, 2017 @08:15PM (#13936387)
The demand for new programs will not abate unless people stop thinking of new things to do with computers. The number of useful things that a computer could do might be infinite -- or it might not. We're unlikely to ever hit that point, because will probably be a long tail that never quite reaches zero.

I don't​ think that programming is a good opportunity for automation, because I think programming is something that humans are actually very good at. In order to write a program, you have to figure out what functionality would be useful to a person, both to elicit latent specifications and to intuit what would be acceptable for things that are not precisely specified. Anticipating what a person would want requires having a theory of mind and understanding the experience of being human. That's about the hardest task in all of computer science. It's also the thing that people are probably best at. On top of just making it work, it has to work well enough be cost-competitive with employing a person.

If we ever get to that point, there will be so few jobs left for humans that almost everyone on earth will be a programmer. So it won't cost very much to employ one.

Re:Programming won’t be automated, or it already has been by defined on Thursday March 23, 2017 @01:20AM (#13937558)
Agree 100%. The fact is (well, in my experience anyway) that the people who want programs written don't really know what they want (they only think they do). The art of requirements analysis is very difficult to do well, partly because of this, and like you say, when (if) any machine becomes intelligent enough to equal a human in eliciting good requirements, we will all be in a lot of trouble.

Right now, is there any natural language system that can appropriately correct the punctuation on these sentences (not meant rhetorically)?

SPLINTERS WEAR SHOES

PANDA EATS, SHOOTS AND LEAVES

Re:Programming won’t be automated, or it already has been by fragmede on Thursday March 23, 2017 @03:10AM (#13937887)
It's impossible, as a human, to know what you actually wanted to say with those sentences in isolation.

Are you setting up for a joke about a gun-toting panda at diner, who left the scene of a crime? Are you paraphrasing an informational sign at a zoo about a black and white animal?

Assuming you're not trying to go for a bit of clever wordplay, however, plug the panda sentence in Google Translate and it'll correctly translate "leaves" to be the plant, and not the verb tense of "to leave", in the chosen destination language.

TechCrunch discussed Google's internal representation of language[1] which implies that while the computer doesn't have a concept of the animal and its food sources, it recognizes that "panda" has something to do with "leaves" as being related to plants rather than "to leave".

Google Translate's far from perfect, but as poor as it is, it was once considered impossible to advance to where it is now.

[1]https://techcrunch.com/2016/11/22/googles-ai-translation-too...

Re:Programming won’t be automated, or it already has been by mtdewcmu on Thursday March 23, 2017 @02:18AM (#13937730)
It's not only cases where the requirements are especially hard to elicit. Computers don't have common sense, so even "obvious" things would be hard.
Re:Programming won’t be automated, or it already has been by TheOtherHobbes on Thursday March 23, 2017 @05:33AM (#13938373)
One reason computers don't have common sense because [insert large but hard to calculate percentage] of human interaction is context, learned and scripted social setting/interaction, and tradition, not semantics. And certainly not statistics.

You can't parse conversations effectively without a social context, because a lot of the meaning is implied by the context. The words themselves are more like pointers into the context, and are never a full description of the meaning. (See also: "Make me a sandwich".)

So when someone on HN says "Time flies like an arrow, fruit flies like a banana" the context tells a human reader that the real meaning is "This is a meta-sentence illustrating a point about grammar, context, ambiguity, and NLP. It does not mean I am saying anything specific about flies, time, arrows, fruit, or bananas."

The contextually correct response would be to continue with a different comment about AI. Or maybe to make a context-bending joke. (But not on HN, where the context doesn't allow that.)

Every so often someone in AI rediscovers this idea and tries to model it, but for some reason it keeps being forgotten. It's been around since at least the 1970s, but so far as I can tell modern statistical ML largely fails to be interested in it.

Re:Programming won’t be automated, or it already has been by mtdewcmu on Thursday March 23, 2017 @01:48PM (#13942161)
There are many tasks related to programming that would be maximally hard for computers. It would be kind of like a gordian knot of all the intractable problems in AI all woven together.
Re:Programming won’t be automated, or it already has been by lr4444lr on Wednesday March 22, 2017 @09:13PM (#13936667)
Very much agree. And I think the corollary, which you describe more than state outright, is that as the tooling and technology gets better, the human value-add would likely demand the talents of the more imaginative, "right-brained", and creatively extroverted - not the stereotypical techie profile. Knuth thought he could do it with literate programming, which didn't take off, to say the least. Maybe it won't ever be that way. Interesting times ahead!
Re:Programming won’t be automated, or it already has been by mtdewcmu on Wednesday March 22, 2017 @10:04PM (#13936897)
Actually, if you consider programs that can be considered art, like video games, that may create an infinite demand for programmers, because the demand for art is infinite. In art, the fact that something has never been done before can be the sole justification for doing it.

With respect to creativity becoming more important relative to technical ability-- I think that both will be important.

Re:Programming won’t be automated, or it already has been by avn2109 on Thursday March 23, 2017 @12:26AM (#13937366)
If the demand for art were infinite, it's a mathematical fact that the price would also be infinite, regardless of the quantity supplied.

In fact the price for most art is very low, below the cost of materials, or free. This implies that demand for art is very, very finite.

Andy Warhol's whole point was that the art that sells for real money has nothing to do with art. It's a sale of association with social status, and the physical art itself is superfluous.

Re:Programming won’t be automated, or it already has been by fjdlwlv on Thursday March 23, 2017 @01:47AM (#13937632)
That's incorrect. Price is limited by capacity to pay. If art production gets more efficient, more people can afford to buy it.

Demand has multiple dimensions. There is only demand for so much quantity of art, but competition can drive the quality up indefinitely.

Re:Programming won’t be automated, or it already has been by mtdewcmu on Thursday March 23, 2017 @02:13AM (#13937714)
I meant infinite in the sense that there will never come a time when the human race stops making art because there's none left to make.
Re:Programming won’t be automated, or it already has been by boomlinde on Thursday March 23, 2017 @03:26AM (#13937928)
The relationship between demand, supply and price that you assume to make that conclusion is not a mathematical fact.
Re:Programming won’t be automated, or it already has been by E6300 on Thursday March 23, 2017 @02:27AM (#13937757)
It would be more correct to say that demand for art is non-zero at every point in time when a person exists. This only becomes infinite if humanity exists for an infinite amount of time.
Re:Programming won’t be automated, or it already has been by laughfactory on Wednesday March 22, 2017 @09:32PM (#13936764)
I agree, and will add that I believe this is the secret sauce in the truly exceptional data scientists, engineers, and developers: people who take the tech skill set and remix with heavy doses of creativity and interpersonal skills.
Re:Programming won’t be automated, or it already has been by elihu on Thursday March 23, 2017 @03:14AM (#13937900)
The demand for new software may never go away entirely, but the scope of what can be accomplished while writing little or no code is ever-increasing. I expect it's going to get harder and harder for businesses to justify spending money on software development when there's already a good-enough program that already exists.

This has been happening for a long time. Back in the 80's and 90's, most of the major tech companies had their own derivative of UNIX because it made sense (sort of) to have their own in-house OS. Now almost everything is Windows, Linux, or MacOS. It would seem very strange for any company to go off and do their own thing rather than use one of those three.

Re:Programming won’t be automated, or it already has been by mtdewcmu on Thursday March 23, 2017 @04:05AM (#13938060)
In the 90s, I thought that there wasn't much software left to write. It turned out that there was. I don't want to be glib, but I think it's always hard to imagine the things that haven't been created yet.
Re:Programming won’t be automated, or it already has been by petra on Thursday March 23, 2017 @04:30AM (#13938149)
I think Moore, after his famous prediction , had some crazy predictions about the stuff computer will do. I would guess Steve jobs had some idea too. So maybe in general , some rare visionaires do know, but most people don't.
Re:Programming won’t be automated, or it already has been by mtdewcmu on Thursday March 23, 2017 @03:42PM (#13943453)
In the case of Steve Jobs, his predictions were things that he did himself. I think what youre saying is that he predicted what a good cell phone would look like. I'm not sure that's really salient to my point.
Re:Programming won’t be automated, or it already has been by TeMPOraL on Thursday March 23, 2017 @06:27AM (#13938600)
> The number of useful things that a computer could do might be infinite -- or it might not. We're unlikely to ever hit that point, because will probably be a long tail that never quite reaches zero.

As it is now, we - as humans with programming skills, not professional programmers employed by few select corporations and their subsidiaries - might never hit that point because we'll be prevented from doing so. Trusted computing, DRM, SaaS model and various other things together conspire to take away the ability to run Turing-complete programs on our own machines. It's something we need to keep in mind too.