Computers

Man-machine collaboration at heart of new Artificial Intelligence XPrize

Man-machine collaboration at heart of new Artificial Intelligence XPrize
he new competition is more open than previous XPrize efforts, with each team able to set its own directions and goals
he new competition is more open than previous XPrize efforts, with each team able to set its own directions and goals
View 1 Image
he new competition is more open than previous XPrize efforts, with each team able to set its own directions and goals
1/1
he new competition is more open than previous XPrize efforts, with each team able to set its own directions and goals

Registration has just opened up for an all new US$5 million XPrize, this time focusing on getting humans collaborating better with artificial intelligence to solve major global issues. Unlike previous competitions, this XPrize, sponsored by IBM's Watson division, doesn't feature a set of pre-determined goals, but instead challenges teams to come up with their own.

You might be familiar with XPrize from its ongoing Google Lunar effort, which is seeing small teams from around the world compete to successfully land a robot on the Moon. It's a seriously ambitious project, and one that has seen rivals team up in the hope of winning out against the competition.

This new project is totally different to the Lunar XPrize, but it's no less ambitious. Aside from the different focus, the big difference here is that the competition is "open," with teams being given the opportunity to pick their own direction. Each participant will be deciding which issues to tackle, and working out how to reach their own goals.

Participants could focus on anything from fixing healthcare or improving education, to taking on major green energy and environmental issues, to any other issue they choose to tackle. XPrize thinks the competition could have a big impact, harnessing the potential of artificial intelligence to solve some of humanity's greatest challenges.

"The IBM Watson AI XPrize will stir innovation and empower a global group of developers, entrepreneurs, and organizations to push the boundaries of human-machine collaboration, forever changing for the better the way in which we live and work." said IBM Watson vice president Stephen Gold.

If you've followed previous XPrize competitions, then the structure of this newest effort will be familiar. A four-year project, prospective participants will have until December 1, 2016 to register, and until March 2017 to submit a detailed development and testing plan.

There will then be three selection rounds conducted each fall, eventually whittling down the competition to ten teams, all of whom will be able to compete for Milestone Prizes – a total of US$500,000 worth of awards to help promising teams continue their research.

At the end of the competition, which will feature three teams taking to the TED2020 stage in front of both online and in-person audiences, a US$3 million grand prize will be awarded to the winning team, $1 million to the runner up, and $500,000 for third place.

Source: XPrize

3 comments
3 comments
Lbrewer42
The problems posted are not really problems for AI. The answer is simple on paper. Get rid of the money-grubbing bureaucracy, hold individuals accountable for their own actions, make jail a place people would HATE to have to be in (hard labor), put a legitimate cap on suing, take the power of omnipotence in money making away from the Fed, go back to money backed by precious metals, and the problems WILL be gone.
But hey, this can only be proven from studying actual, non-media-propagandized history. So let's continue ignore it as always. It will be funny though to see the computers, if truly objective, come up with what has always been staring us square in the face.
Those who refuse to study history are doomed to repeat it.
Truthor
I have an idea that I think should be put in an IBM database along with thousands of ideas by others like me who have an idea but don't want to pursue the work myself. Anyone who wants to work on a project for an XPrize could look over this database for possible project ideas and will have no obligation to give credit for any idea he/she decides to use from this database. Here's my idea: A person's brain knows the honest answer to any clear yes/no question asked of that person (person A). The problem for any other person (B) who wants to know the honest answer to a question B can ask A is how to tap into that knowledge in A's brain. Find a way to do this and we'll have something much better than lie detectors. It might go like this: A sits or lies down and perhaps needs to be hooked up to some computerized equipment or at least be visually monitored by that equipment. A is told not to indicate, vocally or otherwise, any answers to the Y/N questions asked--the answers will occur with YES/NO/I-Don't-Know lights or something similar as revealed by A's own brain. A is not able to prevent the detection of this answer in any way except by not being monitored while being questioned. A further feature may be added if questions may be communicated subliminally. Any non-YES/NO question that B would like to ask could be asked with a request to answer vocally. Then, B may ask a YES/NO question taken from A's answer to the non-YES/NO question, for example if the question was "What town were you born in?", and the answer A gives is "Toledo, Ohio", then B may ask A's brain if A was born in Toledo, Ohio, so at least an honest answer to a non-Y/N question may be verified or a lie detected. A whole project may be needed, if such a device is shown to work with extremely high reliability, to work out the best questions to ask subjects under various situations, courtroom testimony, police investigations, job screenings, political screenings, etc. And a whole bunch of protocols would have to be designed for the most accurate results. For example, definitions of words used in questioning must be made clear to subjects for the best accuracy and fewest 'I-Don't-Know's in using this belief-detection technology. My suggestion for the technology or device name: the 'BOBBI', for 'Brain Operated Binary Belief Indicator'.
Arahant
This is exactly what we need. Bravo for the people behind these competitions. The reality is there are alot of brilliant people out there, if all these people were working on things for the common good the world would be a different place. Problem is people are more likely to want to work for themselves, and I don't really blame them its human nature.... its nature even. So we need proper incentivization for them.
Truthor: I think your post is abit off topic but, the problem with your idea is the nature of whether something is "true" or not is very hard to detect in the brain.. and what it could look like as a brain pattern could be different for each individual or even depending on what the topic(ie brain region) is.
Back to the topic im abit confused as the article says collaborating with AI yet we don't have true AI yet, we just have super computers with some abilities to learn. I think the competition should be towards better developing AI but I'm sure how it is now will be useful aswell, hope some good stuff is developed.