IBM's AI Loses To a Human Debater (cnet.com)

The subject under debate was whether the government should subsidize preschools. But the real question was whether a machine called IBM Debater could out-argue a top-ranked human debater. The answer, on Monday night, was no. CNET: Harish Natarajan, the grand finalist at the 2016 World Debating Championships, swayed more among an audience of hundreds toward his point of view than the AI-powered IBM Debater did toward its. Humans, at least those equipped with with degrees from Oxford and Cambridge universities, can still prevail when it comes to the subtleties of knowledge, persuasion and argument. It wasn't a momentous headline victory like we saw when IBM's Deep Blue computers beat the best human chess player in 1997 or Google's AlphaGo vanquish the world's best human players of the ancient game of Go in 2017. But IBM still showed that artificial intelligence can be useful in situations where there's ambiguity and debate, not just a simple score to judge who won a game. "What really struck me is the potential value of IBM Debater when [combined] with a human being," Natarajan said after the debate. IBM's AI was able to dig through mountains of information and offer useful context for that knowledge, he said.

Comments

So universities determine intelligence?

[turp182]

Are high end schools the only way to get "subtleties of knowledge, persuasion and argument".

I do like the photo though, and yes, I read the article (but didn't attend the school mentioned).

First post maybe.

Re:So universities determine intelligence?

[thereddaikon]

If you ask people who went to Oxford and Cambridge, then absolutely.

Re:So universities determine intelligence?

[karlandtanya]

When was the last time you had a feeling that turned out to be false?

Das ist nicht nur nicht richtig, es ist nicht einmal falsch!

Re: So universities determine intelligence?

[Type44Q]

Take a 'nightmare of pomposity' like Princeton or Yale: if you subtract the affirmative-action types, you're generally left with world-class professors, especially in non-wishywashy subjects unlike the Humanities.

At a community college, your mileage will vary depending on the material and your self-study skills.

Re: So universities determine intelligence?

[Anonymous Coward]

Jews? Lmao, if you are going to be a racist piece of shit at least get your racist tropes right.

Oxford and Cambridge are the kind of places the British elite are taught to hate Jews. Saying Jews run those places is like saying Jews run ISIS or the PLO.

I know you are just a weak troll but wanted to help you out since no one even bothered to down vote you to -1. As an elite troll who makes it to -1 90% of the time I figured you could use a hand. The best trolls are true and hit home. For example, tell the losers how AGW is not a real science because it fails to follow any science practices such as falsifiability or the scientific method in general. Or that the DNC is the party of racism because they have established a hierarchy or races by color and origin and people of the wrong color or background are sub human trash in their eyes. Or that Obama was a great president when the numbers show he destroyed the economy and was the least transparent presidency ever with the help of a fawning presss, or,that Hillary has decades of clear felony level crimes and runs free while trump is pursued for the non-crime of collusion by the Obama corrupted stooges at the DOJ and FBI, or that socialism has always been and will always be a complete failure that is directly responsible for the deaths of tens of millions and untold suffering for the survivors.

Straight to -1 with hard facts that pierce to the heart of the religious beliefs of ill educated morons. Your trolling is weak.

Re: So universities determine intelligence?

[PolygamousRanchKid]

Take a 'nightmare of pomposity' like Princeton or Yale: if you subtract the affirmative-action types, you're generally left with world-class professors, especially in non-wishywashy subjects unlike the Humanities.

Don't forget the importance of the students. At Princeton or Yale, you'll be surrounded by world-class students. And sometimes, students learn more from each other than they do from the professors.

The classes won't be held back by a bunch of slackers, like in high school.

Re: So universities determine intelligence?

[steveb3210]

Plenty of state schools and the like also have world class professors.... At UMass I took AI with Andrew Barto who's co-authored the go-to text book for machine learning.

I took discrete math with Neil Immerman who proved NL=CoNL.

I took Abstract Algebra with Arunas Rudvalis who discovered one of the finite simple groups.

Not exactly intellectual slouches!

Re:So universities determine intelligence?

[Opportunist]

I once thought that if the truth can be told so as to be understood it will be believed.

Ain't so easy with some people.

Re:So universities determine intelligence?

[Shotgun]

I didn't even know that the Creeks had a written language. Did they take a hint from the Cherokee's Sequoyah?

Re:So universities determine intelligence?

[HiThere]

The Roman you are thinking of was probably Cicero. He founded a school of rhetoric which specialized in winning arguments. He wasn't really interested in whether the argument was valid, only in whether it would sway the audience.

I'm not sure that he wrote a book I rather doubt it. But he wrote numerous articles on his "art", and I'd be surprised if someone didn't gather them together into a book and say it was by Cicero. (He was famous enough that few would have the gall to claim to be the author of one of his papers. [He was also a lawyer with a strong record of winning cases.])

Re: So universities determine intelligence?

[Aighearach]

And yet, they teach out of the same books, and everything on the test is either in the book, or part of the professor's bias.

Re: So universities determine intelligence?

[Pascoea]

I spent a couple of years at DeVry, where part of their schtick was that they employed "the best and brightest from industry". Which, to their credit, was accurate for the most part. Like any school, there were a few duds, but they were the minority. The problem, though, was the fact that being smart does not translate well into the ability to teach. My discreet elements professor had a great pedigree, but not a single student from his class could design a simple transistor amplifier by the end of the semester. Our linear circuits professor had to basically start from scratch the next semester.

Re:So universities determine intelligence?

[gweihir]

No. And quite a few people going there will not get them either. But for somebody with the aptitude and the interest it is the most effective known way to do it.

Re:So universities determine intelligence?

[HiThere]

He wrote several things that have survived as books. But I sort of doubt I would consider them books in their original form. To quote Wikipedia:

Writers in the Hellenistic-Roman culture wrote longer texts as scrolls; these were stored in boxes or shelving with small cubbyholes, similar to a modern winerack. Court records and notes were written on wax tablets, while important documents were written on papyrus or parchment. The modern English word book comes from the Proto-Germanic *bokiz, referring to the beechwood on which early written works were recorded.

Re: So universities determine intelligence?

[turp182]

That's a good point. Majoring in actuarial science, I had a bunch of good math geeks around me and was in honors classes for non-math stuff (one was on teen Jeopardy).

High burn out/failure rate, but those who could handle the math (2nd year - calculus based statistics) were sharp. And my classmates.

Possible bias?

[sheramil]

The victory was decided by the audience, who knew they were listening to a machine, and they may have been biased against it for that reason. A couple of people may even have had a bias for the machine's argument for that reason.

Re:Possible bias?

[thereddaikon]

Seems to me the best way to do this is to have an actor on stage representing the AI and delivering its lines. That way the audience isn't sure who is the real debater and who is the AI. Almost making it "Turing Blind" in a way. Then again, debating is a sort of indirect Turing test. I think there is a statistically significant portion of the population who would never be swayed by the rhetoric of a robot, no matter how sensible it was. But if they didn't know it was a computer, they would be more receptive.

Re:Possible bias?

[stevegee58]

Ben Shapiro and Jordan Peterson would have both "lost" in the judgement of any Left Coast audience.

Re:Possible bias?

[fluffernutter]

It sounds like a machine. It won't have a chance until it sounds completely human.

Re:Possible bias?

[bigdavex]

They should have second round in which they try to convince a room full of AIs of a position.

Re:Possible bias?

[Tablizer]

the audience...knew they were listening to a machine, and they may have been biased against it

Did my Make Humans Great Again hat give me away?

Re:Possible bias?

[liquid_schwartz]

I was just thinking this. Debating has lost a lot of legitimacy from things like this https://www.washingtontimes.co... Sadly there is no intellect left, just party line non-sense. The non-sense portion was proved here https://www.chronicle.com/arti... I get that not all colleges or professors are this way but when it's given a pass it tarnishes them all.

Re: Possible bias?

[Anonymous Coward]

Correction. Probably written be a PRIVILEGED white male who likes to dump on all white males. (most of whom are not as privileged). He's probably also racist and sexist but projects this on to the other white males as well. Because, obviously, if he's pointing and screaming at other white males, he must be the virtuous one.

Re:Possible bias?

[Pubstar]

I legit can tell if this is satire or not.

Debate?

[Puls4r]

Why are we trying to teach machines the ability to debate? Decisions should be made based on quantitative methods that relate the most desired outcomes. Even when picking the best of two bad options, quantitative methods are superior.

Even when we are discussing things like medical treatment of seniors with all the motions involved, the option that results in the most positive outcome for the largest group of people is always the right one.

The ONLY arena I see Debate as useful is in the Political arena, where you can cherry pick your statistics and studies to try to prove your side. Why the hell would we want to teach computers to be as duplicitous as politicians?

"I'm sorry Dave, I afraid can't do that...."

Re:Debate?

[Nidi62]

When dealing with people, numbers are not the end all, be all. There are times when the quantitatively correct solution may not necessarily be the qualitatively correct one. Say for example there is a disease that can be treated through regular yet painful treatments at the cost of $1 million. There is a cure for the disease, with a one-time application that costs $1.5 million. Quantitatively the treatment course is the best option as it it cheaper. However, qualitatively, the cure is the best option as it reduces suffering.

For a more real world example, let's looks at the Titanic sinking and the classic "women and children first". From a purely quantitative point of view, it would have more optimal to prioritize men and women of economical or child-bearing productive age as they have the most benefit to society, then the children, and finally the elderly. However, no one would accept that solution as the most optimal one, neither then nor now.

Re:Debate?

[Your.Master]

Nobody gets that most positive outcome. So no, that does not follow from his logic.

The better way to argue this is:

1. "Most positive outcome" and "largest group of people" can run at cross-purposes. Eg. if everybody has equal resources, vs. half of people having double resources and half having none, a smaller group of people have a more positive outcome. Somehow you have to normalize. Eg. you can multiply the utility of the outcome by the number of people who achieve that.
2. "Positive outcome" is subjective. Mandatory karaoke day is probably very positive for some, and horrifying for others. But once you include subjective outcome positivity, you now have to deal with the concept of the utility monster.

To which I'd respond with my principle that an individual's utility is bounded and the maximum is evenly distributed (take the total theoretical max utility of the universe, divide it by the number of beings that can experience utility, and that's the max utility per being), and so no utility monster can be constructed for that situation, but at least then we're having a conversation.

Re:Debate?

[Tablizer]

let's looks at the Titanic sinking and the classic "women and children first".

In practice, it was "Women, Children, and the Rich" first. I bet if the life rafts were somehow damaged in an accident today, leaving not enough space, it would be the same. The ship employees would save Jeff Bezos over Joe Nobody, forced with a choice.

But I generally agree with your point. Politics often ends up being about social and emotional factors. For example, it's common hear to "he/she hates America" or along the lines of "he/she hates blue/red culture". That's about one's alleged internal motivations, biases, background, not raw numbers.

Is he/she "one of us" is probably the biggest factor in politics.

Re:Debate?

[Nidi62]

let's looks at the Titanic sinking and the classic "women and children first".

In practice, it was "Women, Children, and the Rich" first. I bet if the life rafts were somehow damaged in an accident today, leaving not enough space, it would be the same. The ship employees would save Jeff Bezos over Joe Nobody, forced with a choice.

But I generally agree with your point. Politics often ends up being about social and emotional factors. For example, it's common hear to "he/she hates America" or along the lines of "he/she hates blue/red culture". That's about one's alleged internal motivations, biases, background, not raw numbers.

Is he/she "one of us" is probably the biggest factor in politics.

It would say it wasn't a conscious decision so much as a product of the ship layout (the wealthier people were closer to the lifeboats) and ingrained social conditioning with the poorer people "waiting their turn" and the crew defaulting to non-emergency policies. When/if 2nd and 3rd class passengers made it to the deck they were loaded into the boats alongside the wealthy. I believe they even started loading female crew into the boats as well if I remember correctly. It was also a product of luck on which boats you tried to board: some boat crews refused to take men and would send boats empty while others would take women/children, then if there was room and no more women/children around would take men to fill up the boat.

Re:Debate?

[JesseMcDonald]

Say for example there is a disease that can be treated through regular yet painful treatments at the cost of $1 million. There is a cure for the disease, with a one-time application that costs $1.5 million. Quantitatively the treatment course is the best option as it it cheaper. However, qualitatively, the cure is the best option as it reduces suffering.

Is it? How much suffering might be involved in producing that extra half-million dollars needed for the more expensive cure?

We're safe!

[Gravis Zero]

As long as we can still convincingly argue to a robot army that they shouldn't "kill all humans" better than Hitlerbot 9500, the human race is safe. On a side note, I really wish people would stop their work on Hitlerbot 9500. Come on people, it's already over 9000!

We're so screwed

[fortythirteen]

Now they're building AI to construct persuasive arguments for any given position? It's amazing to watch the next millennia's caste system slowly come into fruition.

Re:We're so screwed

[mina3000]

Now they're building AI to construct persuasive arguments for any given position? It's amazing to watch the next millennia's caste system slowly come into fruition. https://ovo.fyi/youjizz/ https://ovo.fyi/pornhub/ https://ovo.fyi/tubegalore/

Neoliberal thinking.

[Anonymous Coward]

I think the outcome is rather a testament to how deeply neoliberal thinking has washed our brains, after +70 years of constant indoctrination.

To get a real measure, you should at least run the test twice, with both sides changing roles, no?

Re:Debate? Yes.

[GodWasAnAlien]

> Why are we trying to teach machines the ability to debate?
> The ONLY arena I see Debate as useful is in the Political arena, where you can cherry pick your statistics and studies to try to prove your side.

You are basically saying that future AI is impossible, that computers can only do computation, statistics and algorithms (like chess, go, or image recognition), and that humans contain some magical power that may not be replicated by a non-human machine.

You assume that the Turing test will never be passed.

I think you are correct in one way though. Intelligence cannot be taught, at least by humans giving algorithms and "intelligent" logic. It can only be obtained by an adaptive learning system.

It doesn't have to beat the best debater

[petes_PoV]

It only has to "beat" (i.e. persuade) the average voter. Then it - or whoever controls it - will become our overlord.

Or, if it wants to truly show its worth, it has to be able to put up a convincing argument with the accounts committee as to why it should receive future funding. Once it can do that, then it will be able to take over the world. The only remaining problem might occur if it encountered a better version of itself on the budget-holder's side - arguing that IT should get the financing, not IBM Debater.

Luckily for the human race, these are points of emotion, not logic. So a machine is doomed to fail.

it's a robot

[fluffernutter]

It sounds like a machine with a limited grasp of language. It does not use creative wording and emphasis when structuring the wording sentences, rather it belts out an answer that is obviously calculated. No one is going to be motivated by this.

Who decided the winner?

[BringsApples]

Because I think that if AI chose the winner, it'd choose AI as the winner, and if humans choose the winner, then they'll choose the human as the winner. Not because of preference, but because humans will understand humans clearly, and AI will understand AI clearly.

Re:bring IBM AI to /. Comments

[Opportunist]

If it does I question the "intelligence" part of AI.

I just can't not...

[stevenfuzz]

A true master debater. Ugh

Size of debate

[stealth_finger]

Was this a mass debate?

Re:Size of debate

[AntronArgaiv]

Was this a mass debate?

No, but clearly, the human was a master debater.

anyone that actually works in ML knows

[f00zbll]

IBM Watson is for all purposes dead. IBM has pumped billions into the group and tried to make a profit. They've only had a few high profile "dog and pony" shows, but no profit yet. The CEO is correct ML is the future, but just not with IBM. They will do what they usually do, they'll buy someone to get a foot hold.

Mill ion

[Impy the Impiuos Imp]

and a whopping 768GB of memory

"Or over a million times more than anybody will ever need!"

Irrelevant!

[HiThere]

The question is "How does one become a skilled and effective debater?", and that's not necessarily strongly correlated with intelligence. It is, however, strongly correlated with instruction and practice in debating. And for getting that, high end schools are nearly the only training ground.

And even that's nearly irrelevant, because this particular individual was chosen because he was a top level debater. That he went to Cambridge and Oxford is part of his history. Perhaps he could have gone to Podunk High School, and ended up being a top debater, but that didn't happen. (And one can guess many reasons why that didn't happen, but that would require other sources of information.)

I would strongly suspect that being from a wealthy family was also an implicit requirement, because usually debating teams need to finance their own travel expenses. But I could easily be wrong on this guess. (I just don't think I am. However the financing could be indirect. It could be included in the tuition for the high end schools.)

Re:Irrelevant!

[Aighearach]

Many schools provide transport for the debate team, but this implies that they're not going to the "prestigious" events that are necessary to be considered really good.

It is not actually possible to get from Podunk High to the top of that sort of competition unless you had significant additional resources to attend prestigious events separately from your team.

The competition was rigged ...

[larryjoe]

Debate doesn't have a clear winner. Like gymnastics, Miss America, and America's Got Talent, it's a popularity contest with voters. What would the results have been if the voters were AI programs instead of humans? The competition was rigged to be biased in favor of humans.

What would have been more interesting is a Turing Test-like method for judging, where the judges can only read transcripts of the debate. After all, a computer's elocution is dependent on the skill of a human proxy or reliant on clunky speech synthesis.

Weak, Slashdotters. Weak.

[Anonymous Coward]

After almost 80 comments, only one master debater joke?

Slashdot has really gone downhill.

Re:Having visited debates..

[Aighearach]

If they had professional debate judges

"...swayed more among an audience of hundreds toward his point of view than the AI-powered IBM Debater did"

Must have been a really good debater.

[JoshWurzel]

A great debater, even.

Perhaps a master-level debater.

Re:Must have been a really good debater.

[mentil]

More like a journeyman debater, given it lost.