Trust is the missing link in the successful implementation of self-driving cars. Just as with electric vehicles, it will probably take years before trust is earned. But unlike an electric car, autonomous ones (AV) will have a lot more work to do before reaching that point.
There are a lot of reasons why humans fear AVs, some rooted in reality, other in a visceral feeling that something is not quite right. After all, after more than a century of seeing humans in control of the car, how can one easily accept a vehicle moving on its own?
For DARPA one of the main reasons trust is lacking is the technology’s inability to be judge and juror for itself. Regardless of how smart they are, none of the systems currently available can evaluate their own abilities and admit when something is beyond their scope.
In an attempt to fix this issue, DARPA announced the launch of the Competency-Aware Machine Learning (CAML) program. It is aimed at making machine learning systems capable of continuously assessing their own performance, and most importantly communicate to whoever is listening when they believe they're not up for a task.
“If the machine can say, ‘I do well in these conditions, but I don’t have a lot of experience in those conditions,’ that will allow a better human-machine teaming,” said Jiangying Zhou, a program manager in DARPA’s Defense Sciences Office. “The partner then can make a more informed choice.”
Of course, all the back and forth communication between man and machine will be a bit more practical than that. If the CAML program is successful, at the end of it AVs will be able to say how many times they did something and with what accuracy, giving the human a much more informed decision to make.
DARPA is awaiting potential contributors to enlist in the program by February 20. Those willing to take part need to have knowledge in the fields of AI, machine learning, pattern recognition and much more. Full details on the program can be found in the document attached below.
For DARPA one of the main reasons trust is lacking is the technology’s inability to be judge and juror for itself. Regardless of how smart they are, none of the systems currently available can evaluate their own abilities and admit when something is beyond their scope.
In an attempt to fix this issue, DARPA announced the launch of the Competency-Aware Machine Learning (CAML) program. It is aimed at making machine learning systems capable of continuously assessing their own performance, and most importantly communicate to whoever is listening when they believe they're not up for a task.
“If the machine can say, ‘I do well in these conditions, but I don’t have a lot of experience in those conditions,’ that will allow a better human-machine teaming,” said Jiangying Zhou, a program manager in DARPA’s Defense Sciences Office. “The partner then can make a more informed choice.”
Of course, all the back and forth communication between man and machine will be a bit more practical than that. If the CAML program is successful, at the end of it AVs will be able to say how many times they did something and with what accuracy, giving the human a much more informed decision to make.
DARPA is awaiting potential contributors to enlist in the program by February 20. Those willing to take part need to have knowledge in the fields of AI, machine learning, pattern recognition and much more. Full details on the program can be found in the document attached below.