AAI and GWAP

When Amazon.com introduced their Mechanical Turk service they also gave the underlying principle a name: Artificial Artificial Intelligence (AAI).

Artificial Artificial Intelligence (AAI) is a term coined by Jeff Bezos with reference to Artificial Intelligence (AI). Certain processing tasks, such as identifying whether a person in a photograph is male or female, are still performed better and faster by humans than computers. AI is not yet adequate to programming such tasks. The idea of AAI is to outsource those parts of a computer program to humans.

(Wikipedia – Jeff Bezos)

Recently, Philip E. Ross wrote about AAI as used at Pandora in his IEEE Spectrum article Want Advice? Try an Expert:

If you want recommendations and you’d rather not rely exclusively on your customers, you can always do the daring thing and consult actual experts. That’s the idea behind Pandora, a free Internet radio service that employs musicians to rate songs according to a checklist of criteria, such as pace, rhythm, even the voice of the performer.
[...]
It may seem strange to use so much manpower as a supplement to computer power, but it makes sense when humans alone can handle the job—a peculiar field sometimes called artificial artificial intelligence. One example of AAI is setting puzzles, or “captchas,” for visitors to a Web site to solve, both to prove that they’re human beings and not bots and to perform some useful chore, such as deciphering the blurred letters from a scan of an old book. Other AAI programs lure people to do such work by providing entertainment or, as Amazon’s Mechanical Turk does, money.

I think it might make sense to think of games with a purpose (GWAP) as a subset of AAI. Research has shown that popular GWAP games can attract very productive players. Those applications have the strong appeal of direct and instant gratification: Entertainment. This allows for a powerful productivity factor: Intrinsic motivation. People want to play and they want improve on their scores or beat other people’s scores.

I wonder in how far it might make sense to ponder a game approach for some AAI applications not just to get people to do work for (nearly) free, but also to improve results quality, if the people providing the data are actually paid to do so.

games with a purpose – introductory notes

There are a number of problems that are very hard for computers to solve, but comparatively easy for humans. One example of this is image classification: It presents a tough challenge for computers, but humans are typically easily able to point out individual objects in a photograph as well as their locations.

Luis von Ahn coined the expression game with a purpose. The idea here is that it would be very expensive to pay people to do certain tasks, particularly if those tasks involve large volumes of data. Engaging games on the other hand provide the immediate incentive of entertainment. Games with a purpose are played for the sake of enjoyment, but have the side effect of getting real work done by the players, too.

The following three papers provide a useful introduction.

• Luis von Ahn, Games with a Purpose [PDF], IEEE Computer, June 2006. Introduces the basic concept, briefly outlines some of the games in development as part of von Ahn’s research and looks at future applications.
• Luis von Ahn and Laura Dabbish, Designing Games with a Purpose [PDF], Communications of the ACM, August 2008. Presents a discussion of general design principles for games with a purpose. Luis von Ahn’s games are available here.
• Katharina Siorpaes and Martin Hepp, Games with a Purpose for the Semantic Web [PDF], IEEE Intelligent Systems, May/June 2008. Applies GWAP principles to build useful information ontologies. The games are available here.

Here is a clip of Luis von Ahn discussing some of his games and the concept of human computation.

I have no doubt that we will see lots of interesting applications of these principles in coming years.