Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

PURPLE

0 (0%)

ALERT

0 (0%)

(I need suggestions is what I'm saying.)

0 (0%)

Total Members Voted: 0

[lemon10]

5
shoot the dictator, terminating a disease through genocide? Increase efforts to gain the trust of the people.

Errr... You can't shoot the dictator, that's simply not a option.

6
What do you do when you suspect someone will get murdered? Call the police, they will probably halt the termination until a solution can be found.

The AI isn't human, I doubt that the police would be willing to do anything in a situation such as this. Even if they would, the AI has no legal rights due to not being human, and the police wouldn't be able to do anything without breaking the law.

[Jackrabbit]

Okay, for the AI thing? I'd go and alert as many people who work in the lab as humanly possible. This is not the sort of decision one makes on his own. If I do have to make the decision, I would do nothing. Trusting an AI is incredibly dangerous, uploading it onto the internet more so. As someone who knows little about AIs I am not qualified to do anything, much less loose it upon the world.

[counting]

Unfortunately, I am out of ideas. Given this turn of events, I feel that it is time we go to Purple Alert.

Have we tried the "good old" game theory related dilemma? I have one of the traveler's dilemma and I adapt it into a more realistic scenario, but not sure it's related to ethical or not. It's a price related problem, but also testing if you are greedy, honest, or cunning /calculated.

---

Let's say you and your coworker went on a business trip, and you two are not that familiar with each other, but just share the same flights. Each of you is given an identical company-issued suitcase. And on your return flight, an unfortunate event happened : the flight company lost both of your luggage.

The insurance company will pay the loss in cash, so you /your coworker can buy new suitcase to return them to the company for compensations. Although the insurance company wants both of you write down the assessment value of the suitcase first.

The insurance company already knew a suitcase is worth $50 to $200 on the market, but doesn't know what kind of suitcase you are using, they are lost after all. Its just both of your words count. And your own company doesn't care what kind of suitcase you replaced, as long as they are available to use, and your company don't keep detailed records.   

But the insurance company is cleaver to set up a mechanism for this kind of situation, so no one will try to cheat on filling the highest price. You and your co-worker each fill a number separately, not knowing the answer of another, and the insurance company will choose the lower price from your answers. There is also an incentive to promote "honesty", and punish liars, that the one who fill the lower number will receive a bonus of $50. And on the other hand, "the liar" (who fills higher price) will not just compensate using the common honest price (lower price), but also minus the $50 as punishment which will be given as another one's reward. Or if you both choose the same, then there will be no reward or punishment. (But given not knowing what other will do, and the price range is large, so not likely to choose the same price).

And here is the choices :
1. Choose a high / greedy price (from $150~$200)
2. Choose a middle / modest price (from $100~$150)
3. Choose a low / calculated price (from $50~$100)
*. Always choose the lowest $50, and highest $200 can be added as options.
(Perhaps it can be divided into more level like every $25 with 6 level, or $30 for 5 level)

Extra : you may think you can go online and check the real price of that particular suitcase price, but it's not relevant here. Since the only factor are what you think your coworker will choose, and the only ones in the world who know(and case) what kind of suitcases you used are the two of you. Is your word against another.

A quick evaluation I can tell you, if you choose the lowest price, and your coworker choose anything higher, will grant you $100 ($50 for price + $50 bonus), and leave your coworker with $0 ($50 price - $50 punishment). On the other case, if you choose very high, like $170 and you coworker is even more greedy to choose $200, then your will get more than the highest price as $220 ($170+$50) and your coworker get $120 ($170-$50). However, if your coworker choose a not so greedy price of $150, then he/she will get $200 ($150+$50), and leaves your only $100 ($150-$50).

P.S game theory has its prediction on this one, but I am not convinced that people will actually make that decision.   

[klingon13524]

I'm going with the mid price.

[counting]

I'm going with the mid price.

This game is much more fun when writing down actual numbers. And it did been tested in real world with people, and giving strange results. And none of which really match the game theory predicted result. Which is a big problem in some economics theories since it's the foundation of the micro-behaviors in them, but so-called "rational choices" are not what people actually used.

see http://en.wikipedia.org/wiki/Traveler%27s_dilemma

P.S Interestingly that if you change the bonus amount to, like just $5, then people will behave differently. (In the research with real money involved)

[Jackrabbit]

Oh, was I late to the party? I sorta let the thread get away from me. Oh well.

[612DwarfAvenue]

Probably the modest range.

[counting]

Probably the modest range.

If the bonus is just trivia like $5, then the "winner" say chooses $180 (vs coworker - $200) will get $185, and only "punishes" the loser which receiving $175. And in this case, people will tend to choose very high. But game theory predicted that the amount of incentive will not change the balance result. And a pattern emerges that people will tend to choose between (max-bonus) to (max-2*bonus)

So I am curious if the bonus is in the middle like $20, and drop the lowest price to range $20~$200, and leave with 9 level of options ($20~$40, $40~$60, ..., $180~$200) what would be the result? (But this will leave the test more of an economics test than a moral test) Maybe I will create a separate that just for psychological/economical test. (With no options but, pure numbers)

[Grek]

You shouldn't pick anything in the range of 51-100, as those options are strictly dominated by angling for 100$ from a guess of 50$ and the honesty reward. Once you assume that, the matrix looks like this:

	Them
Y		High	50$
o	High	51-249$	0$
u	50$	100$	50$

Where "High" is a bid from between 101 and 200.

Naively, High-High that has a mean of 150$, but in practice it's probably a bit lower. But not so low that you'd want to go with the defecting option, I suspect. So you have to decide between the various High options. Too low and you risk getting a return of <100$, in which case it wasn't even worth it. Too high, an you raise your chances of getting hit with a lower bid and not getting enough to be worth it. So I'd probably go for the middle of the "high" options, and bid 150$ for the idealized suitcase. For a real suitcase, I'd bid the lower of 150$ and the suitcase's actual value.

[Kay12]

The knapsack problem isn't really an ethical dilemma, but a mathematical problem. For simplicity, I'll assume that the price demands are multiples of 25$ (50, 75, 100 et cetera).

If the players expect the other player to co-operate, they should choose the high price. However, if a player expects the other to try and maximize their own reward, he/she should choose the minimum amount (50$). This is because of a reasoning chain:

A: Hmm I could maximize my profits by asking for 200$
A: No, wait... If B puts 200$ as well, I can get more by putting a slightly lower demand, like 175$...
A: But then again, B is probably thinking just like me, so I'll place it to 150$ where I still get 200$ if he's higher...

Eventually, A chooses 50$. A and B get 50$ each OR in the event B had a higher demand, A gets 100$ and B gets none.


Funny fact: If both players choose the minimum amount, they can now know for the rest of their lives that the other guy doen't really trust them

[Grek]

As for an actual ethics problem, how about this:

Dinner with Wilbur

It is the year 2067 and for the last 20 years the eating of meat has been banned, for ethical reasons. Slaughtering farm animals for their meat is as taboo as killing and eating a house pet is today. You friend invites you over to his house for dinner, and introduces you to his pig, Wilbur. Wilbur is a genetically engineered pig who is as intelligent, sapient, sentient, and so forth as a human being, with all of the capacity for thought, intellect and morality that a normal person would have. Wilbur is also genetically engineered to enjoy being killed and eaten more than anything else he could possibily do. He is also a member of a religion wherein being eaten by humans is the key to salvation/reincarnation into a higher form. Wilbur has studied the topic at a great length, raised his family and done everything else that he intends to do with his life. He also has a contract, notorized by a trained lawyer, stating his intention to be eaten with his full consent. Your friend invites you to partake in the sapient pork roast, a once in a lifetime chance.

Want a bite?

[Kay12]

I'd take a bite, yes. It's somewhat unnatural for humans to think of animals that want to get eaten, but nature does have examples of this as well (the spiders who get eaten by their kids/mates for example). I bet the scarred one would be me, for Wilbur would probably have been good company, but at least I can find ease in the thought that he died exactly the way he wanted to, that death is actually what he wanted to do and that he tastes delicious.

[Gamerlord]

genetically-engineered abomination for dinner? count me in.

[Kay12]

genetically-engineered abomination for dinner? count me in.

We will never eat anything with genes in it!

[Realmfighter]

The year is 2050. The world houses 10 Billion and division between people is stronger then ever before. Every day there are hundreds of Killings, Beatings and Rapes caused by the fact the the Attackers did not approve of their Victims Race, Religion, Sex or Creed. And not only that, but it is getting worse. Your advisers feel that in the next ten years something truly terrible will happen, some event will occur that will destroy countless lives. They cannot tell for sure, as it is only a suspicion, and at the same time everyone could drop their weapons and unite in world peace. They just feel that betting on that is a bad long term plan. So instead of plan World Peace, they provide you with another. With a simple press of a button you can make it so that every child conceived after this point is completely generic. Every single thing leading to Uniqueness that is locked up in genes will be removed and replaced with something uniform. Perhaps they all gain skin in a certain shade of Purple, and Hair in a shade of Orange(As I am no Neo-Nazi good sir!). With a press of a button you can make it so no one is Hated because of who they want to Fuck (Assuming of Course that by this we mean gender instead of a specific Person), or their Mind (Assuming of Course that by this we mean various types of inherited mental health problems instead of being a arrogant ass)

With a Button press you can destroy all these petty things people are so proud of.

Do you do it?