Peter Norvig: The machine age

http://www.nypost.com/p/news/opinion/opedcolumnists/the_machine_age_tM7xPAv4pI4JslK0M1JtxI

by pitdesi • 15 years ago

95 52 15 years ago

l0nwlf2• 15 years ago

``One of the biggest issues in AI is determining what the question is really asking i.e. translating a “natural language” query into something machine can understand and find the appropriate answer.'' - so True.

jbri6• 15 years ago

That's also one of the biggest issues in software development.

celoyd2• 15 years ago

Surely it’s the issue in software development. Everything else is a case or subproblem of “how do I use a computer to solve this imperfectly defined problem?”

I mean, it’s borderline tautological, isn’t it? Software development is the practice of translating natural-language requests into executables. That’s not just a recurring question – it’s definitional.

This is why I’m nervous when strongly antisocial people say they want to go into software. The point of this field is to figure out what people need (or want, or mean, or like). Of course you can do this creatively rather than responsively, and with more rather than less respect for your customers. But you don’t escape human tastes and concerns when your job is to explain them to a microprocessor.

bluekeybox• 15 years ago

One would be mistaken to think that this problem is trivial even to us humans.

Luyt• 15 years ago

A link to the 'print version' of the page: the whole article on one page without the incredible clutter and cruft that the New York Post splashes all over its site:

http://www.nypost.com/f/print/news/opinion/opedcolumnists/th...

Smerity• 15 years ago

In this AI/machine learning is compared to the space race and the fight against cancer - I like the comparison considering I work in the field ;) Unfortunately though I think AI and machine learning has the fundamental problem of rolling improvement...

Think back a hundred years to a time before the internet, long distance telephones, common air travel and so on. To people from a hundred years ago we'd be the most outlandish science fiction story ever told.

I consider the same thing whilst watching science fiction films - in reality they wouldn't be amazed that they can reach Mars in a half hour. They'd instead be complaining about the in-flight entertainment or the turbulence caused by navigating asteroid fields.

AI has the same problem. People can now (naively) communicate with each other even though they share no common language. Texts and documents can be re-written in real time to be legible in another language. Credit card fraud and spam emails are handled transparently by systems trained on hundreds or thousands of hours of cumulative knowledge. No-one sees these things. No-one respects these things.

Human kind judges itself from the position it currently stands, not from where it began. For this reason alone incremental improvements will never really be considered milestones of achievement.

zeteo• 15 years ago

"To people from a hundred years ago we'd be the most outlandish science fiction story ever told."

You'd be surprised:

http://en.wikipedia.org/wiki/Paris_in_the_Twentieth_Century#...

derleth• 15 years ago

Technology is one thing; society is another: I doubt anyone from the 19th Century could possibly imagine a debate over gay marriage. A crazy promoting it getting beaten down by a righteous world, possibly, but not a debate.

Cushman• 15 years ago

I don't know if the comparison is that inapt; how many people do you think are really aware of the technology that goes into making that little box say "in five hundred yards, turn left"?

mattdeboard• 15 years ago

This is a great perspective. I would add that people are less impressed by Watson not just because it's an incremental step, but because they don't understand the principles and advances Watson is designed to showcase.

From what I gather about the coverage of Watson in traditional media, this seems to be billed as, "Can a machine out-think a human? Tune in at 7:30pm to find out!" To most modern people, this is kind of a yawner. Of course a machine can "out-think" (read: think faster than) a human. It wouldn't be very exciting if the news came on and talked excitedly about a computer that could perform long division faster than a human, either.

But that's not really what Watson represents, is it? The importance of Watson is something that doesn't register with most lay people: The processing of natural language into 'machine readable' data, then querying an enormous database to find relevant info, then choosing the best answer.... "Isn't that what Google does? Like, millions of times a day? What's the big deal?"

(Disclaimer: Most people have better things to worry about than advancess in AI and ML, so I'm not saying it's a bad thing they don't understand what makes Watson special "under the hood." And to be fair, I was way more impressed by Word Lens than I am by Watson.)

tl;dr: The concepts that make Watson so impressive aren't easily discernible from what already exists in the world, for most people.

mthomas• 15 years ago

Perhaps offtopic, but I find it odd that there is an opinion piece by Peter Norvig in the NY Post.

pgbovine• 15 years ago

true, but i'm sure that NY Post has more subscribers than NY Times, so if his goal was to reach the largest possible audience, perhaps NY Post was an appropriate venue

necubi• 15 years ago

This isn't true. According to Wikipedia, the NY Times has 876,638 daily/1,352,358 Sunday subscribers [0] while the NY Post has 525,000/343,361 [1]

[0] http://en.wikipedia.org/wiki/NY_Times [1] http://en.wikipedia.org/wiki/NY_Post

LostInTheWoods2• 15 years ago

Not to be snobbish, but the NY Post is considered a working class paper. Usually bought by those more interested in the funnies, movie times, and local sports, than say international affairs or ramblings about AI.

bmh1002• 15 years ago

This is an excellent article for raising awareness of AI. However, I disagree with its conclusion that we will always relate to AI as tools, pets, advisers, etc.

When we ask whether machines can think, I believe it is a question of volition and self-direction. Does the machine have an open-ended goal and is self aware? Can it alter its own programming to change its own methods, its own knowledge, and its own goals? If the answer to these two is "yes," then I believe we will consider the machine to be thinking.

What are the implications? If machines gain those capacities, as well as the ability to compute emotions, then what separates them from us? If you agree that humanity is fundamentally a function of our minds, then, if machines can compute in fundamentally equivalent ways, they become "human." What evidence is there that this is forever impossible?

tgflynn10• 15 years ago

If you agree that humanity is fundamentally a function of our minds

What about the ability to feel pleasure and pain ?

I don't believe a deterministic machine will ever experience such subjective feelings (though it may behave as if it does). Computers may well become sapient but they (at least with current technology) will never be sentient.

jbri6• 15 years ago

What is "pleasure" and "pain"? Ultimately, they're just electrical signals triggering the release of certain chemicals in the brain, with subsequent alterations to how the brain processes other stimuli.

"It doesn't really feel pain, it just looks like it does" is probably how many people justified atrocities against other humans they viewed as somehow "lesser".

feral3• 15 years ago

Thought experiment: If there was a human that suffered an injury that left them unable to feel pleasure, or pain - some form of nerve damage, perhaps - I wouldn't think any less of that human's mental function. I wouldn't think they had lost any humanity, or argue they should be accorded any less human rights. I think that they'd still have hopes, and fears, would be much more important than their ability to feel pain or pleasure.

Given this, how important is the ability to feel pleasure, or pain, really?

I think you are focusing on the wrong thing, and disagree with this way of assessing humanity.

adrianN7• 15 years ago

You are basically arguing about P-Zombies[1]. I think that line of argument is fallacious.

What if the computer were powerful enough to perfectly simulate the workings of a human brain? Does that brain not have a consciousness?

[1] http://en.wikipedia.org/wiki/P-Zombie

wooby• 15 years ago

The episode "The Thinking Machine" of the Nova miniseries "The Machine That Changed the World" is my favorite introduction to the history of AI up until around AI winter.

The episode, along with four others, are available here: http://waxy.org/2008/06/the_machine_that_changed_the_world_t...

Jun8• 15 years ago

"Learning turned out to be more important than knowing." This seemingly simple sentence, to me, sums up the article, i.e. the data driven approach as opposed to model-driven ones.

mtrn• 15 years ago

Single page (print) version: http://www.nypost.com/f/print/news/opinion/opedcolumnists/th...

TeMPOraL• 15 years ago

"Spam filtering programs using A.I. learning and classification techniques correctly identify over 99.9% of the 200 billion spam e-mails sent each day."

200 billion spam e-mails each day. Something should be done with spammers just on the basis of how much electricity it wasted on sending, processing and discarding all those e-mails...

Peter Norvig: The machine age

Related Stories