Sentient world: war games on the grandest scale

Sim Strife

By Mark Baard • Get more from this author

Posted in Government, 23rd June 2007 09:02 GMT

Free whitepaper – Enabling The Agile Data Center

Alok Chaturvedi wants SWS to match every person on the planet, one-to-one.

Right now, the 62 simulated nations in SEAS depict humans as composites, at a 100-to-1 ratio.

One organisation has achieved a one-to-one level of granularity for its simulations, according to Chaturvedi: the US Army, which is using SEAS to identify potential recruits.

Chaturvedi insists his goal for SWS is to have a depersonalised likeness for each individual, rather than an immediately identifiable duplicate. If your town census records your birthdate, job title, and whether you own a dog, SWS will generate what Chaturvedi calls a "like someone" with the same stats, but not the same name.

Of course, government agencies and corporations can add to SWS whatever personally-identifiable information they choose from their own databases, and for their own purposes.

And with consumers already giving up their personal information regularly to websites such as MySpace and Twitter, it is not a stretch to imagine SWS doing the same thing.

"There may be hooks through which individuals may voluntarily contribute information to SWS," Chaturvedi said.

SEAS bases its AI "thinking" on the theories of cognitive psychologists and the work of Princeton University professor Daniel Kahneman, one of the fathers of behavioural economics.

Chaturvedi, as do many AR developers, also cites the work of positive psychology guru Martin Seligman (known, too, for his concept of "learned hopelessness") as an influence on SEAS human behaviour models. The Simulex website says, if a bit vaguely, SEAS similarly incorporates predictive models based upon production, marketing, finance and other fields.

But SWS may never be smart enough to anticipate every possibility, or predict how people will react under stress, said Philip Lieberman, professor of cognitive and linguistic studies at Brown University.

"Experts make 'correct' decisions under time pressure and extreme stress that are not necessarily optimum but work," said Lieberman, who nevertheless said the simulations might be useful for anticipating some scenarios.

JFCOM's Blank agreed that SWS, which is using computers and code to do cultural anthropology, does not include any "hard science at this point".

"Ultimately," said Blank, "the guy to make decision is the commander." ®