Slashdot Mirror
New IBM Robot Holds Its Own In a Debate With a Human (nbcnews.com)
PolygamousRanchKid shares a report: The human brain may be the ultimate super computer, but artificial intelligence is catching up so fast, it can now hold a substantive debate with a human, according to audience feedback. IBM's Project Debater made its public debut in San Francisco Monday afternoon, where it squared off against Noa Ovadia, the 2016 Israeli debate champion and in a second debate, Dan Zafrir, a nationally renowned debater in Israel. The AI is the latest grand challenge from IBM, which previously created Deep Blue, technology that beat chess champion Garry Kasparov and Watson, which bested humans on the game show Jeopardy.
In its first public outing, Project Debater turned out to be a formidable opponent, scanning the hundreds of millions of newspaper and journal articles in its memory to quickly synthesize an argument on a topic and position it was assigned on the spot. "Project Debater could be the ultimate fact-based sounding board without the bias that often comes from humans," said Arvind Krishna, director of IBM Research. An audience survey taken before and after each debate found that Project Debater better enriched the audience's knowledge as it argued in favor of subsidies for space exploration and in favor of telemedicine, but that the human debaters did a better job delivering their speeches.
The AI isn't trained on topics -- it's trained on the art of debate. For the most part, Project Debater spoke in natural language, choosing the same words and sentence structures as a native English speaker. It even dropped the odd joke, but with the expected robotic delivery. IBM's engineers know the AI isn't perfect. Just like humans, it makes mistakes and at times, repeats itself. However, the company believes it could have a broad impact in the future as people now have to be more skeptical as they sort out fact and fiction. "Project Debater must adapt to human rationale and propose lines of argument that people can follow," Krishna said in a blog post. "In debate, AI must learn to navigate our messy, unstructured human world as it is -- not by using a pre-defined set of rules, as in a board game."
In its first public outing, Project Debater turned out to be a formidable opponent, scanning the hundreds of millions of newspaper and journal articles in its memory to quickly synthesize an argument on a topic and position it was assigned on the spot. "Project Debater could be the ultimate fact-based sounding board without the bias that often comes from humans," said Arvind Krishna, director of IBM Research. An audience survey taken before and after each debate found that Project Debater better enriched the audience's knowledge as it argued in favor of subsidies for space exploration and in favor of telemedicine, but that the human debaters did a better job delivering their speeches.
The AI isn't trained on topics -- it's trained on the art of debate. For the most part, Project Debater spoke in natural language, choosing the same words and sentence structures as a native English speaker. It even dropped the odd joke, but with the expected robotic delivery. IBM's engineers know the AI isn't perfect. Just like humans, it makes mistakes and at times, repeats itself. However, the company believes it could have a broad impact in the future as people now have to be more skeptical as they sort out fact and fiction. "Project Debater must adapt to human rationale and propose lines of argument that people can follow," Krishna said in a blog post. "In debate, AI must learn to navigate our messy, unstructured human world as it is -- not by using a pre-defined set of rules, as in a board game."
Completely fake. The topics were prearranged, and yes they were "assigned on the spot" but there was a predetermined list. IBM is desperately trying to sell their AI snakeoil. If AI worked, why not have it solve REAL problems that people will pay for, rather than parlor tricks like plying Go, and Chess and other games?
So, it scans human-generated content, and then builds a plausible sounding argument to support whatever position you give it.
This thing is going to cause a lot of unemployment in politics.
This is a cool project, but the article is utterly useless without a transcript.
"First they came for the slanderers and i said nothing."
If this AI truly uses real facts in a debate it would be wonderful. One thing most "debaters" these days seem to despise is actual facts. They get in the way of an emotional argument, something I (sadly) see as most prevalent in the SJW crowd. They have nice-sounding ideas that appeal to emotion but do not stand up in the face of factual examination.
This is also going to derail politicians in a big way, especially if it sticks to facts. Politicians hate facts. They bank on their constituents not knowing the facts and being too lazy to check them.
In the end they will lay their freedom at our feet and say to us, Make us your slaves, but feed us. - Fyodor Dostoyevsky
One could plausibly make the same assertion about humans. There is nothing magic about the brain, after all.
File under 'M' for 'Manic ranting'
There is no "AI" on this planet and this thing is just a collection of dumb reflexes that give the appearance of an intelligent agent.
And so are you.
Of course, you can argue otherwise, but then the AI can make such an argument as well, and will likely do a better job at it (according to the article).
So how should I determine which stream of electronic communication was generated by "an intelligent agent" and which one by "a collection of dumb reflexes"?
The most logical conclusion is that there's not much of a difference.
The "AI debater" mainly seems to search for possibly relevant statements in a large library and then inject them into the debate. Throwing factoids at each other is clearly how debates happen these days take place and how many "decision makers" operate.
But that isn't how debates ought to take place. Debates should start with premises and mutually agreed facts and then reach conclusions via reason and logic.
In its first public outing, Project Debater turned out to be a formidable opponent, scanning the hundreds of millions of newspaper and journal articles in its memory to quickly synthesize an argument on a topic and position it was assigned on the spot. "Project Debater could be the ultimate fact-based sounding board without the bias that often comes from humans," said Arvind Krishna
If the data it uses to "argue" comes from human sources, it has a human bias.
That being said, it is cool technology and it demonstrates how bad human debate can be. If you can win an argument without actually knowing what you are talking about (which you can), it demonstrates the (lack of) value debate can have; it also underscores the lack of real value in the level of political discourse that we have today. We spend a lot of time arguing over things we don't really know about.
No, it didn't.
Yes, it did.
No ....
Have gnu, will travel.
>> that isn't how debates ought to take place
I disagree. This was clearly an even match between two master debaters.
Let me know when the computer can win a Slashdot debate. There's no way it could cope with this sort of argument:
Computer: "AI has made great improvements in it's cognitive ability."
Anonymous Coward: "Yeah, WELL FUCK YOU!!!!"
AC wins every time.
If you post as Anonymous Coward, don't expect a reply.
So IBM is claiming this can be used as a fact-based sounding board, but if it is looking through published work how does it know that what the system is repeating is actually fact? I realize that humans have this same issue, but if you are going to present your device as a paragon of factual information, then I would expect a rigorous system of validation to be part of it.
I will say being able to build this type of language structure in a way that is at least passable is quite an achievement in and of itself. I have the feeling "holding it's own" is an overstatement, but it was apparently not ridiculous.
Not a rhetorical question. Like all the pseudo-intelligent software being trotted out the last several years, it does not know how to 'think'. Try asking it "What are you doing right now, and why are you doing it?" and let's see what it says. All this software is doing is sifting and sorting information, and arranging it into statements, and it doesn't matter if the statements it's making are in response to statements made by the human debater, the machine does not understand what it's doing, just like it doesn't understand anything at all; there's no mind in there, it doesn't 'think', it just processes information, and it's not relevant so far as I'm concerned that it happens to do that in a sophisticated and remarkable way. Not impressed, it's just another dog-and-pony-show to placate investors and stockholders.
IBM's engineers know the AI isn't perfect.
I detest writing like the above. People trot out the "I know I/it/whatever isn't perfect" lead-in all the time, and I dislike it because it's seductively-packaged idiocy... it costs the speaker nothing (who would, or even could, argue that xyz IS in fact perfect?), while then paving the way for them to follow with an equally vapid statement that does nothing to inform.
I was recently trying to assess whether buying expensive retainers for my son's post-braces teeth would be worthwhile, and asked his orthodontist what the success/stability rate with them was. She replied, "Well, we can't guarantee perfection, of course, but most people like them." Which cornered me into "being rude" by explaining to her that the fact that the outcomes weren't "perfect" was not informative or helpful; do they work in 80% of patients? 99? 40? THAT information is helpful.
- First they ignore you, then they laugh at you, then ???, then profit.
'And it would lean'?
--- whatever it was programmed to 'trust'.
So if it trusted sites primarily from Iran it would learn the 'Koran' is the only word of God.
If it trust sites primarily from the Chinese government it would learn 'there is no God'
if it trusted sites from the united states is would learn that 'Christianity brings hope' but 'whatever you believe is ok' and would stop debating completely ! ( * LOL * )
âoeTolerance applies only to persons, but never to truth. Intolerance applies only to truth, but never to persons.
Not knowledge, however this thing can process vast amounts of knowledge and provide a response which has a very high probability of being correct and/or at the very least relevant, and it can do this in real-time.
If users could follow directions regarding computers, or even have a two-way conversation with this machine, then this thing could take over a third of my duties in I.T. (Handholding, and transliterating). As it stands the machine may reduce my time to research answers, a problem which Google has largely solved.
I suppose this could provide a high level of research to those who need to learn how to research, making them ineffective at their jobs without technology, and unable to accurately access the quality of the technology leading to stagnation and the collapse of modern society.
Deploying this thing in a field of debate shows an effective alternative to reducing the value of intelligence. If we were to capitalize on this for educational and testing purposes, somehow outlawing and blocking its use in commercial environments, it could prove an incredible asset for building a brighter future. We could potentially use this machine to sharpen our brains rather than to dull them.
It is a sad state that we equate Debate skills with leadership skills.
Debates are something you need to win or loose. Not an open discussion to grow and learn. You can win a debate on a false idea or lie over someone who has the truth and data on their side, however they may lack the debate skills to try to convince a neutral party. Often the best and well thoughout idea is far more complex then what can be stated in quick sound blurbs.
Presidential debates over the past few generations have not been really productive. Most of us are already had made up their mind on who they are voting for, most will just vote for whoever has a R or D on their party affiliation regardless of their stance. So the Debators neutral party is just a tiny fraction of the population. And for the most part they are trying to read non-verbal queues. (such as Nixon sweating) or finding someone just loosing their temper. The topic up for debate are not relevant as we have a good idea where the stance is.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
If you're proposing a magical, non-physical component to the human mind, then isn't the the onus on you to provide support for your extraordinary claim?
--- Most topics have many sides worth arguing, allow me to take one opposite you.
This platform is not a presence in a debate as we think of it. It has no inherent values, no physical experience, and it doesn't have anything at stake other than the demo itself. It is not free of bias, rather, it sorts through the information and biases humans supply it with.
That last part is why it's still super-valuable, in my opinion. In my time as a politically engaged citizen of the USA, there have been times when the big media across the political spectrum has lagged for almost a year in reporting information that could be found with very good corroborating evidence in major media across the globe. Regarding the Iraq war, major stories seemed to break in the Summer or Fall after the invasion, but they were only breaking stories in the USA. They'd been reported on extensively before the invasion in Europe in Lebanon. I'd been reading that media and cross-comparing and detecting US corporate bias, noticing what was either left out or buried in the footnotes, and becoming aware of the biases and motives they implied.
One really stunning example: word came out that the BBC and CNN were both carrying "full" transcripts of Hans Blix's testimony to the UN about the efficacy of UN inspectors to verify Iraqi compliance with denuclearization, but that CNN had omitted major parts of a "full" transcript. I did the homework. I downloaded both transcripts and broke out my tools as a Linux guy and analyst, and did the diff. What amounted to two large paragraphs on my screen right in the middle of the testimony were the omitted parts. They were the most detailed and convincing parts of Hans Blix's testimony, and the most relevant to a public that had a right to informed participation about whether the nation should start a pre-emptive war. That a "liberal" institution doctored verified and significant news in favor of a pro-war stance was really, really damning.
That wasn't the end, that story goes on.
Point is, as a human, cross-comparing many diverse pieces of information and journalism has definitely brought not just the story, but how some actors are trying to manipulate the story to light. We need an equitable, fairly administered system to make this sort of analysis available to the public. It needs to detect discrepancies and focus the public on where it can validate and verify something into being closer to fully true. It needs to be broad-based enough to not be itself a prop for those looking to use it for propaganda.
I'm all for using this AI in ways that might help critical thought prevail.
So, for example, discussing the news in the UK, where most of the newspapers have a mainstream bias, this poor AI will just parrot the same old rubbish you can read in papers such as the Telegraph, Times and Guardian, or worse the Mail, Express or the Sun. Also it is wrong to associate what is in those papers with facts. All these papers bend and distort, over report or omit in order to fit their agenda. Rubbish in, rubbish out.
I don't think there would be a winner in a Trump vs machine debate. Trumps "wins" in debates come from undercutting arguments with completely fabricated bullshit, and if that doesn't work he'll start throwing insults to throw the opponent off their game. That track just won't work with a machine, because they wouldn't have the emotional base to get upset enough to lose their cool.
I too would like to see this debate. The machine *MIGHT* release the blue smoke from the never ending stream of made-up day-dreams, but it'd still be fun to watch the first few rounds as Trump was confronted with actual facts.
think of how many call center employees just answer simple questions out of a database. That's what this is for. Parts of India & the Philippians are genuinely worried about the job loses.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
Guys, this is really silly. As Godel has already demonstrated, it is impossible for a machine to meet the criteria of consciousness. "Artificial intelligence" is a chimerical idea and is not possible.
"Imitative intelligence" would be more accurate. A machine may be able to hold a facade of "intelligence," but any semblance of intelligence has been derived from its creators.
The claim that the machine "synthesized an argument" is misleading. Machines are not capable of a priori. The machine simply sorted information giving the appearance of a synthesized argument. The author projected this activity of synthesizing an argument onto the machine, but that is not what happened.
Then the author of the article made the incredible claim that the machine does not have bias, but just the same, they fed it a junk-food diet of newspaper articles & mental garbage.
This article is propaganda.
They're trying to persuade you to believe that machines can be intelligent, that machines will soon be just as or more capable than men at thinking, and that human mental faculties are mechanical. Perhaps the hope is that the general populace will eventually fall under of a large "appeal to authority, or argumentum ad verecundiam" umbrella and give up critical thinking altogether. This is already happening to people in STEM, who have largely ignored philosophy, and evidently cannot think rightly.
Feed the AI garbage data and it won't know the difference. It won't feel puzzled or shameful or anything. Humans can at least second guess the "truth" that is given to them. An AI is happy to drive into a wall without giving it any thought, because that was the "most logical conclusion"
Feed a newborn human garbage data and it won't know the difference either.
Oh, you meant an adult human? One that had their neural network constantly trained with real-world data for a couple of decades?
Well then, shouldn't an AI compete on a level playing field?
While every healthy human has some intelligence, they also have free will and can chose to not use that intelligence. So doing what some humans do does not prove intelligence. A bit more is required. This machine is impressive, no doubt, but it does not have intelligence of any kind. It has zero understanding of what it does. That, incidentally, is also true for many humans.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Just ask this machine to debate something not discussed in the press. It will immediately fall flat on its face. I will not.
But since you are a physicalist, you do not have active general intelligence anyways (it being a fundamentalist quasi-religious belief at this time), so arguments from you are worthless.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Some humans can. It requires a member of the 10% or so of independent thinkers, the rest cannot. They will eat up whatever garbage is fed to them. So while we do not and will not anytime soon have intelligent machines, we are finding out that the average human is not using most of the intelligence it has available. Not really a surprise.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Not knowledge, however this thing can process vast amounts of knowledge and provide a response which has a very high probability of being correct and/or at the very least relevant, and it can do this in real-time.
It actually processes data, not knowledge (yes, the distinction is critical here) and it cannot actually do this in real-time. It takes significant pre-computation time to get there and only the last step is somewhat real-time. But that is not the main problem. It unfortunately has no "very high probability of being correct" (as it does not do any fact-checking and there are countless things the press consistently get wrong) and being "relevant" without being correct is worse than having nothing. This is were anti-vaxxers and flat-earthers come from. Their arguments are most decidedly relevant, but they are deeply wrong.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.