Could Google ’s artificial intelligence arrangement be impertinent enough to get a business at Google ? It ’s sure not insufferable .
Google Deepmindhas rise an algorithm that is able to resolve a notoriously knotty “ 100 - hat riddle . ” The enigma requires such high-pitched level of lateral thinking and problem solving that it ’s been present during interviews for investment banking company Goldman Sachs and , ironically , Google .
Here ’s how the conundrum goes :
“ An public executioner draw up 100 prisoners single file and puts a loss or a blue hat on each prisoner ’s head . Every captive can see the chapeau of the people in front of him in the line - but not his own hat , nor those of anyone behind him . [ There is also an unknown number of red or blue hat ]
The executioner starts at the end ( back ) and asks the last prisoner the color of his lid . He must do “ violent ” or “ blue . ” If he suffice correctly , he is allowed to survive . If he render the incorrect answer , he is pour down instantly and silently . While everyone hears the answer , no one have sex whether an answer was right .
On the night before the line - up , the prisoner confer on a strategy to help them . What should they do ? ”
Within the web , each of the 100 prisoners was modeled as a disjoined independent agent . However , to find a solution they must together with sour together and transmit .
Image acknowledgment : Jakob Foerster et al .
There is an optimal solution in which you’re able to 100 per centum hold open 99 of the prisoners , with the remaining one prisoner having a 50/50 chance .
The Francis Scott Key is to make a protocol that establishes if there ’s an odd or even number of one color of hat . For example , the first prisoner in the queue could say “ blue ” to mean there is an even identification number of drab hats in front of them , or “ red ” to intend there is an funny act of blue hats . From this , remaining captive can then work out their hat colour from the number of leftover and even chapeau left that they see in front of them and the responses they ’ve heard behind them .
If you ’re still confused – do n’t worry . The solution ’s complexity further illustrates the acute floor of data point that these " rich neuronal internet " can treat ; a level of data so high it can create the same consequence as a chemical group of humans during a fiddly and creative trouble - clear bodily process .
“ It ’s basically a first step toward have AIs that can communicate and join forces , ” articulate Jakob Foerster , who worked on the research , said toNew Scientist .
He added , “ They ’ve come up with protocol that are dissimilar from how human being solve these problems .
“ We do n’t yet in full understand what the root are , but we know that they play . ”
you could read the full studyhere .
