Fascinating story, this. IBM’s Project Debater AI system debated itself onstage over whether or not AI posed a threat. The debate occurred on November 21, 2019, at the Cambridge Union in the UK.* The result: “AI will do more good than harm” was the winning side.
Here’s a sketch of how the debate worked: “Speech-by-crowd technology makes it possible to collect free-text arguments from audiences and automatically deliver persuasive viewpoints whether to support or argue against a topic.” Supposedly, according to Noam Slonim, an IBM engineer, this tech could link decision makers and the people they impact by their decisions.
Maybe. I guess we could use all the input we can get before we make decisions. But it’s a bit of a mob-rule thing, or should I say, mob-rule think? “Project Debater digests massive texts, constructs a well-structured speech on a given topic, delivers it with clarity and purpose, and rebuts its opponent. Eventually, Project Debater will help people reason by providing compelling, evidence-based arguments and limiting the influence of emotion, bias, or ambiguity.” That is, unless it receives fudged evidence and falsified information.
Therein lies the problem with almost any AI at this time. And it also lacks the key to most debates: Winning the emotion and not the mind of the audience. Let’s assume that AI can deliver the tightest of logical arguments. Haven’t we witnessed numerous times when logic is deemed flawless, but humans decide to do the irrational that sometimes turns out to be a good decision?
Decision-making is often hit-and-miss. Whereas it is true that getting numerous ideas from numerous people CAN provide us with a larger base of knowledge and perspective, it is also true that numerous ideas from numerous people CAN also distort knowledge and perspective. Yes, I agree with you that I should gather as much information and as many different perspectives as I can before I make any decision, but like you, I’m finite. Of necessity, I make many decisions on the fly. I make none omnisciently and none with knowledge of the unexpected. Unintended consequences will probably accompany all of us throughout our lives, even if we have access to Project Debater AI and an audience of the most astute people in the world to provide possible perspectives for our self-debating process.
So, in what circumstances would you yield to the arguments of AI? In matters of great importance to a society? In matters of great importance to you personally? But what of those decisions that are six-of-one and half-a-dozen of the other? What of those decisions that are merely a matter of preference? And what of the time element for your finite existence? Do you really have time to allow an AI to digest the thoughts of a crowd to be submitted and processed before you make your decisions? In times of urgency, I think you might tend to act alone.
But, AI aside, aren’t we living in an Age of Speech by Crowd? Aren’t we inundated daily with so many voices we choose to apply the filters of bias and preference? That’s the way our decision-making often goes, and it often works out all right when we act without crowd influence. Any crowd can become a mob, even an “intellectual” mob. Remember, one of the reasons that America’s Founding Fathers decided on a democratic republic rather than on a pure democracy is that the passion of a mob can switch with gossip. Just because a majority group-thinks doesn’t guarantee that the majority’s decision is the best decision.
What would have happened if Harry Truman had used AI before he dropped the bombs on Hiroshima and Nagasaki? What about any president making any decision on the world stage. Time to wait for AI to gather as many sides as possible to resolve the issues in a debate? Maybe the decisions you make don’t have consequences for thousands or millions of others, but on the international stage some do have to make such decisions. And that’s one of the problems that has to be addressed in developing AI: How do ethics (or morality) figure into a debate and the final decision? Using AI? How does one decide how to decide?
At the end of that onstage debate at the Cambridge Union, the audience voted on the winner of the debate. Note that. The audience voted. Think all those people in the audience were perfectly rational in assessing the winner? If so, then why did some vote for one side (51.22%) whereas others voted for the other side (48.17%), and some abstained (0.61%)?
*Cohen, Nancy. 26 Nov 2019. AI debate machine argues with itself at Cambridge Union. TechXplore. https://techxplore.com/news/2019-11-ai-debate-machine-cambridge-union.html Accessed November 26, 2019.
Here’s a sketch of how the debate worked: “Speech-by-crowd technology makes it possible to collect free-text arguments from audiences and automatically deliver persuasive viewpoints whether to support or argue against a topic.” Supposedly, according to Noam Slonim, an IBM engineer, this tech could link decision makers and the people they impact by their decisions.
Maybe. I guess we could use all the input we can get before we make decisions. But it’s a bit of a mob-rule thing, or should I say, mob-rule think? “Project Debater digests massive texts, constructs a well-structured speech on a given topic, delivers it with clarity and purpose, and rebuts its opponent. Eventually, Project Debater will help people reason by providing compelling, evidence-based arguments and limiting the influence of emotion, bias, or ambiguity.” That is, unless it receives fudged evidence and falsified information.
Therein lies the problem with almost any AI at this time. And it also lacks the key to most debates: Winning the emotion and not the mind of the audience. Let’s assume that AI can deliver the tightest of logical arguments. Haven’t we witnessed numerous times when logic is deemed flawless, but humans decide to do the irrational that sometimes turns out to be a good decision?
Decision-making is often hit-and-miss. Whereas it is true that getting numerous ideas from numerous people CAN provide us with a larger base of knowledge and perspective, it is also true that numerous ideas from numerous people CAN also distort knowledge and perspective. Yes, I agree with you that I should gather as much information and as many different perspectives as I can before I make any decision, but like you, I’m finite. Of necessity, I make many decisions on the fly. I make none omnisciently and none with knowledge of the unexpected. Unintended consequences will probably accompany all of us throughout our lives, even if we have access to Project Debater AI and an audience of the most astute people in the world to provide possible perspectives for our self-debating process.
So, in what circumstances would you yield to the arguments of AI? In matters of great importance to a society? In matters of great importance to you personally? But what of those decisions that are six-of-one and half-a-dozen of the other? What of those decisions that are merely a matter of preference? And what of the time element for your finite existence? Do you really have time to allow an AI to digest the thoughts of a crowd to be submitted and processed before you make your decisions? In times of urgency, I think you might tend to act alone.
But, AI aside, aren’t we living in an Age of Speech by Crowd? Aren’t we inundated daily with so many voices we choose to apply the filters of bias and preference? That’s the way our decision-making often goes, and it often works out all right when we act without crowd influence. Any crowd can become a mob, even an “intellectual” mob. Remember, one of the reasons that America’s Founding Fathers decided on a democratic republic rather than on a pure democracy is that the passion of a mob can switch with gossip. Just because a majority group-thinks doesn’t guarantee that the majority’s decision is the best decision.
What would have happened if Harry Truman had used AI before he dropped the bombs on Hiroshima and Nagasaki? What about any president making any decision on the world stage. Time to wait for AI to gather as many sides as possible to resolve the issues in a debate? Maybe the decisions you make don’t have consequences for thousands or millions of others, but on the international stage some do have to make such decisions. And that’s one of the problems that has to be addressed in developing AI: How do ethics (or morality) figure into a debate and the final decision? Using AI? How does one decide how to decide?
At the end of that onstage debate at the Cambridge Union, the audience voted on the winner of the debate. Note that. The audience voted. Think all those people in the audience were perfectly rational in assessing the winner? If so, then why did some vote for one side (51.22%) whereas others voted for the other side (48.17%), and some abstained (0.61%)?
*Cohen, Nancy. 26 Nov 2019. AI debate machine argues with itself at Cambridge Union. TechXplore. https://techxplore.com/news/2019-11-ai-debate-machine-cambridge-union.html Accessed November 26, 2019.
RSS Feed