AI vs. human debate

Author: Tejretics

Posts

Total: 8
Tejretics
Tejretics's avatar
Debates: 9
Posts: 501
3
4
8
Tejretics's avatar
Tejretics
3
4
8

Barney
Barney's avatar
Debates: 53
Posts: 3,466
5
9
10
Barney's avatar
Barney
5
9
10
Very cool!

I will however point to the bias of the setup. The human had a potential 79 points to gain, of which he gained 17. The machine had only a potential 21 points to gain.
Smithereens
Smithereens's avatar
Debates: 2
Posts: 502
2
2
4
Smithereens's avatar
Smithereens
2
2
4
-->
@Barney
The side which is more compelling should theoretically never have less people agreeing with it after the debate than before. 

RationalMadman
RationalMadman's avatar
Debates: 574
Posts: 19,931
10
11
11
RationalMadman's avatar
RationalMadman
10
11
11
@Smithereens

The side which is more compelling can be the side debated worse. This is because people can be utter idiots and also because a genius debater vs a terrible debate can't convince you that something like morality is objective since that's utter bollocks but may win the debate hands down and deserve no loss of points for taking a blatantly false side and murking his opponent in spite of it.
Barney
Barney's avatar
Debates: 53
Posts: 3,466
5
9
10
Barney's avatar
Barney
5
9
10
-->
@Smithereens
I assume that theory depends upon a perfect audience to grade it, and more compelling being objective instead of subjective?

The thing I really wish we knew, was if indeed the winner lost any of their original believers.
Smithereens
Smithereens's avatar
Debates: 2
Posts: 502
2
2
4
Smithereens's avatar
Smithereens
2
2
4
-->
@Barney
I think it strikes a good compromise between using an unqualified set of judges and having a meaningful judgement. I've actually never considered using an aggregate statistic for before/after positions in an audience to judge a debate. That way they don't have to explain to the audience how to judge a debate, they can just find out who was more persuasive overall. But yes I agree it's a far from perfect system and there's a few nuances that I'd like to know too. 
3RU7AL
3RU7AL's avatar
Debates: 3
Posts: 14,582
3
4
9
3RU7AL's avatar
3RU7AL
3
4
9
-->
@Smithereens
I think it strikes a good compromise between using an unqualified set of judges and having a meaningful judgement. I've actually never considered using an aggregate statistic for before/after positions in an audience to judge a debate. That way they don't have to explain to the audience how to judge a debate, they can just find out who was more persuasive overall. But yes I agree it's a far from perfect system and there's a few nuances that I'd like to know too. 
Intelligence Squared debates use this method and get played on the radio. [LINK]

And sometimes on television. [LINK]

10 days later

Tejretics
Tejretics's avatar
Debates: 9
Posts: 501
3
4
8
Tejretics's avatar
Tejretics
3
4
8
-->
@Barney
I think this setup is flawed for other reasons, but net gain is gains - losses; in other words, not losing people, for the purposes of this set up, is counted as a gain. So both sides had a potential 100 points to gain, for all intents and purposes.