Veritasium

We need your help on an upcoming video!

Here is an infamous paradox. What would you do? ve42.co/NewcombParadox

We need to record your answers (anonymously) in the Google form, so please click through and answer there.

Thanks so much!

6 months ago | [YT] | 5,800



@christianschafer3724

The real paradox: I have to sign in to answer anonymously.

6 months ago | 427

@ThatItalianMapper

Double it and give it to the next person

6 months ago | 1,000

@khclupus

Take the mystery box. Either you get $1,000,000, or you get to be the guy the computer was wrong about. Also I’m fortunate enough that the $1,000 won’t make a noticeable difference in my life, but the $1,000,000 will.

6 months ago | 790

@k26898

I love that we, the audience, get to be part of these! And then we can see our contribution reflected in the video :)

6 months ago | 554  

@SergioGugliandolo

I take both: my cat likes empty boxes

6 months ago | 46

@kamilmrowka7983

I feel like it might be a similiar situation to the experiment with two buttons. One button has a 60% change for a reward, 40% for a punishment. The other one has a 40% chance for a reward, 60% for a punishment. In that experiment, humans tend to try to game the probability and never really reach the 60% rate of success. Simpler animals like pidgeons learn that pressing the first button results in rewards happening more often. And after some learning time, they alway pick the 60% button and perform better than humans. In this experiment, we know that the AI is almost 100% accurate, and we know that pretty much all of the people who chose to take two boxes, walked out with $1000, and pretty much all of the people who chose just the mystery box, walked out with a milion. And even though in theory the predictions were already made. The AI is extremely good and if it's almost never wrong, why trying to game it, don't you think it would have thought that that's going to be your choice? Pick the single box, it's almost 100% that it predicted it. I understand the argument of, "the boxes were already chosen", so there's no downsides. Cause what you do now can't change it. But If I know that the predictor is so accurate, I'm just going to trust it predicts everything, and assume my thoughts are not outsmarting it, and it was able to pretty much trace them step by step. I'm picking just the mystery box. I'm looking forward to the video. I love paradoxes. On one hand I feel like I'm being logical here. On the other, this way of thinking requires you to make some weird assumptions, but an almost perfect predictor AI is a weird assumption, so If it's a part of the question, I'm just sticking with it being so accurate. Paradoxes like this one usually turn into an endless argument between: - But the AI is so accurate - But the boxes are already filled And the answer kind of depends on which one of these you find more important.

6 months ago (edited) | 7

@DennysTheRestaurant

I take the computer because it would help me make more than a million dollars

6 months ago | 869

@bcunningham3718

I flip a coin, it can’t predict me if I can’t predict myself

6 months ago | 104

@ish_

By completing the form, I’m training the ai to put 1 million into the box

6 months ago | 19

@RudeusGreyrat9

Guys the person behind the predictions is Michael (Vsauce), he's trapped in the supercomputer!

6 months ago | 76

@zapking5

It is unclear to me if the person in the question is meant to know about the prediction, and whether the AI knows that the person would know.

6 months ago | 548

@jerryqu4998

I take just the mystery box. Because I am that kind of person, and the supercomputer knows it!

6 months ago | 163

@tiusic

It's a weird causality problem. By the time you are choosing which boxes to take, the money is already in the boxes, so straightforwardly there's no reason not to take both. But then naturally that's what you were predicted to do. Essentially it depends whether you think that your decision in the present can influence the prediction in the past, which is some very funky causality.

6 months ago | 464

@taufiqutomo

Newcomb's paradox is more a test of faith than a test of logic. Those who have faith in the predictor only pick box B. Those who don't have faith in the predictor pick both boxes. So, I'm afraid it matters who makes the prediction.

6 months ago (edited) | 66

@comparatorclock

I feel like the survey needs a second part for the prediction reveal and then a second question about what you pick

6 months ago | 15

@jeronstevens8491

So when do I get the million dollars?

6 months ago | 210

@bigpopakap

Thank you for allowi me to clearly establish a pattern of choosing the mystery box, so when this happens to me, the supercomputer will know how effing quickly and surely I selected "only mystery box", so it will know that that will be my choice. By establishing my intent clearly, I've made it the correct choice! 🎉

6 months ago | 8

@mitulbhatnagar983

My thought: The outcome depends on the prediction from this god machine and not my choice. I might at least have $1000 😂

6 months ago | 90

@Villianase

i’m taking the million dollars this would be much harder if the amount of money was closer together and i can’t get over my bias

6 months ago | 2

@caxve

In that situation, the outcome had already been decided. If theres a million in the box, take both the boxes and get 1.001 million. If there isnt, take both and get 1000. Nothing you do can get the million back. Therefore take both. If you knew beforehand though then take the single box. Because your choice then changes the outcome.

6 months ago | 24