Urgent need for AI

Theory of games is simple and elegant mathematical approach for modeling human behavior. I have used it to model the decision of humanity when confronted with questions like: is global warming real or not and, is asteroid impact possible or not? When you look at the discussions flaming everywhere from Facebook to G+, you can see many confronted opinions. Using Gambit, we can make very simple strategy game confronting “Humanity” with an extinction level event. Not surprisingly, humanity would take course of action leading to salvation only if cost for such action would not exceed certain threshold. Simply put: if salvation is not too expensive, then yes, we would eventually choose a strategy which is trying to prevent such disastrous event. If the price is too high, humanity chooses simply to deny existence of possibility of such an event.

Lets start with first setup when “humanity” thinks that salvation comes under the “right price”. There are two players in this game (actually three when we take into account “chance” or “mother nature” player). One of the players is humanity (red) with 100 units value at their disposal. ELE (extinction level event) eradicates entire value and the branches are showing moves or courses of action taken by the players. Fictions “salvation company” (blue) is an entity that can “invent” solution which can prevent such event, but not for sure – let’s say that they have 50/50 chance for success. Now, in order to save the world, “humanity” has to give away certain amount of it’s value to the salvation company. If the price is 25 units (or quarter of entire population value), part of humanity would be willing to give away their possessions to the “salvation company” even under assumption that they will not be always successful in the effort to find the solution.

The flow goes from left to right: first, a threat is detected (lets say, global warming). Now, humanity can deny or seek solution. If they deny, that costs 0 and salvation company gets 0 in any case. However, there is 50% chance (or 1/2) that such event is destructive at global level and results with loss of entire property, in which cases payout for humanity is -100. If it is non destructive, payout is 0. On the average, payout for this option is -50 for humanity (given that we have chance to play such game more than once).

The other strategy emerging in this setup is to “pay” to “salvation company” a quarter of value (indicated with loss of -25). “Salvation company” gains 25 units whether successful or not. However, this branch gives lower loss in longer course of actions (remember, we are trying to play this game more than once). All in all, the game predicts two Nash equilibria (small table below) where humanity is taking two separate courses of action – one to deny and other to seek solution.

Now, here is the same game, but this time with price higher that 1/4 of value that should be given to “salvation company”. This time humanity has to give away HALF of its value in order to seek for the solution. (This is not that hard to imagine, because if fictitiously we would need to give away use of cars entirely and some other CO2 emitters, that would be pretty much as giving away half of perceived value in someones life.) Now, the decisions change dramatically. The only remaining dominant strategy is denial. That means, that even if all scientists are right about the grave future consequences of CO2 emission and global warming, humanity in its whole would not accept seeking solution as viable option. Just take a look at the simulation:


What can we conclude from this? Sooner or later we are going to be confronted with extinction level event created by us, or by chance alone. Theory of games is predicting that humanity even under such extreme conditions is not willing to give away it’s rational attitude (read: selfishness). If we leave decision to us alone, we will make grave mistake and conduct wrong decision (we deny). That is the main reason why we need good artificial intelligence right now in this age. It is not something that could just  make our lives easier, artificial intelligence could literary save us from obliteration caused by our own blindness. It could be the fourth player in the game that could change the tip of the tide in our favor…

Source: http://artificialintelligencebooks.glegleme.com/post/Urgent-need-for-AI.aspx


About basicrulesoflife

Year 1935. Interests: Contemporary society problems, quality of life, happiness, understanding and changing ourselves - everything based on scientific evidence.
This entry was posted in Are We doomed?, Common. Bookmark the permalink.

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.