Computers Plus Crowds Could Tackle World's Toughest Problems
by Charles Q. Choi, Live Science Contributor | December 31, 2015 04:27pm ET
|
Credit: Sergey Nivens/Shutterstock.com
|
The world's most dire problems, such as climate change and global
conflicts, could be solved using a combination of human and computer
intelligence, researchers say.
Human outperform machines at many tasks, such as recognizing images and
thinking creatively. So, with the help of computers, crowds of people
could collaborate in networks to achieve what neither people nor
computers could do alone, a growing field known as
human computation.
"What's most exciting to me about human computation is that it gives us
hope today," said Pietro Michelucci, director of the Human Computation
Institute in Fairfax, Virginia. Although many people have pinned both
their hopes on
artificial intelligence (AI), or super-intelligent machines, human computation provides an alternate view, he said.
By using today's technology to combine humans and machines, human
computation could achieve sooner what AI might achieve only in the
distant future, Michelucci said. And, "with the integral involvement of
humans in these systems as both participants and stakeholders, we can
better ensure that we remain in control," he said.
One notable example of human computation is reCAPTCHA, an online widget
used by about 100 million people daily when they transcribe distorted
text into a box to prove they are human in order to access online
content
.
This act of transcribing collections of letters has helped power
efforts that have digitally transcribed 13 million articles from The New
York Times archives.
Most of today's human-computation systems rely on doling out small
"micro tasks" to many people and then merging the results together. For
instance, 165,000 volunteers in 145 countries have used the EyeWire
platform to analyze thousands of images online and help build the
world's most complete map of the neurons in
the human retina, which is the tissue in the back of the eye that detects light and enables people to see.
However, as effective as micro tasking has proven to be, this strategy
alone cannot address so-called "wicked problems" such as climate change
and global conflicts, experts said. [
10 Technologies That Will Transform Your Life]
"Wicked problems are wicked because they have many interacting parts
[and] unpredictability, and because we don't understand how the
different parts feed back on each other," said Janis Dickinson,
professor and director of citizen science at the Cornell Lab of
Ornithology in Ithaca, New York. Michelucci and Dickinson analyzed the
latest research in human computation in an article published in the Jan.
1 issue of the journal Science.
And trying to solve wicked problems can have unforeseen and unwanted
consequences — for instance, giving financial aid to a country after a
natural disaster can lead to corruption that can actually stymie relief efforts, the researchers said.
Now scientists are envisioning ways in which human computation might tackle such complex problems.
"The key to addressing wicked problems is to create a working model,
[a] computer simulation, of all of the interacting systems that pertain
to a given problem," Michelucci told Live Science. "Imagine something
like the game SimCity, but a thousand times more detailed. Then link in
real-time sensors attached to the
Internet.
The more faithful the model is to the real world, the more accurate it
will be for testing out solutions and predicting outcomes."
Imagine an online system that feeds this working model of the world
"with knowledge from real people, where a doctor can input diagnostic
methods, a mechanic can describe how a piston works, and farmers in
every region of the world can provide local updates about
agricultural pests,"
Michelucci said. "A working model of the world that pristine requires
working knowledge that may be spread across the minds of thousands or
millions of people, books, electronic documents and data sets."
This strategy for tackling wicked problems requires not only the
constant gathering of data from the real world, but also the use of
multistep reasoning. Under this method, each problem gets broken down or
"decomposed" to many simpler parts that are easier to address.
New human-computation technologies might help make this a reality;
recent techniques allow contributions from people to get processed by a
computer and then sent to others for improvement or analysis of a
different kind, the researchers noted. [
9 Odd Ways Your Tech Devices May Injure You]
For instance, YardMap.org was launched in 2012 to map global
conservation efforts one parcel of land at a time, and it allows
participants to interact and build on each other's work, something that
crowdsourcing alone cannot achieve. Other examples of multistep
reasoning were seen in the Polymath Project, which helped prove an
80-year-old mathematical theorem, and the ePluribus Problem Solver,
which generated a factually accurate and well-constructed journalistic
article based on just a handful of photographs. In both cases, diverse
participants worked together to generate solutions.
Creating a working model of the world to handle wicked problems also
requires creative thought in order to see how wicked problems might
evolve in response to attempted solutions, Michelucci said.
"We can draw on human computation
methods for stimulating innovation,
eliciting new ideas, spreading them around and giving people the
opportunity to build on each other's work," Michelucci said. "Of course,
all this has to be fun, easy and quick, so that millions of people
actually choose to participate."
"The first step might be to elicit broad solution classes from human participants, such as
halting climate change or adapting to it," Michelucci said.
"Then, each of those [solution classes] might be further delegated to
humans for decomposition — 100 people might receive the task of
decomposing 'halt climate change' into two subclasses, such as
'biological solutions' and 'physical solutions.' Each proposal is then
sent by the computer to 100 more people who evaluate it on various
dimensions.
Then, each of these ideas would be sent out to 100 more people, who
might break them down further or propose specific solutions, like 'paint
our roofs white to reflect sunlight back into the atmosphere.'
"Ideas would then propagate through the system through various stages
of vetting and modification," Michelucci said. At any stage, experts
could step in to help explain complex problems in plain English.
Michelucci and Dickinson noted that human computation will need many
improvements before it could tackle wicked problems. For example, in
most human-computation efforts, only a small number of participants do
most of the work, Michelucci and Dickinson said, adding that researchers
want to find ways to maximize recruitment and contributions of
participants.
"There are many questions about how people behave in human-computation
systems that must be resolved before we can think really big about their
use in humanitarian efforts or disasters or monitoring and addressing
problems arising with chronic
environmental change,"
Dickinson told Live Science. Moreover, Michelucci and Dickinson
cautioned that researchers needed to consider what human computation may
mean for the labor force, unemployment rates, and the economy, so that
people who contribute to human computation projects are protected from
exploitation.
But
crowdsourced efforts such
as Wikipedia and crowdfunding platforms such as Kickstarter highlight
the massive potential that human cooperation has for solving problems,
Dickinson said.
"There are huge social benefits to cooperation that have largely been
overlooked — think of reputation and reciprocation or lack thereof,"
Dickinson said. "By providing the right kinds of information about our
cooperative efforts and where we stand as cooperators, human-computation
systems can provide unprecedented support for people to help work on
large problems that require large-scale human effort to solve."
Follow Charles Q. Choi on Twitter @cqchoi. Follow us @livescience, Facebook & Google+. Original article on Live Science.
A mushroom cloud from the world's first successful
