The US Defense Advanced Research Projects Agency (DARPA) recently announced its Competency-Aware Machine Learning (CAML) program, which aims to “develop machine learning systems that continuously assess their own performance in time-critical, dynamic situations and communicate that information to human team-members in an easily understood format.” The idea behind this is that once machines can assess their own competence, it should be easier for humans to trust them, which will greatly facilitate cooperation.
As Jiangying Zhou of DARPA explained: “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,” because the human partner will then be able to “make a more informed choice.” While the program eventually aims to develop machine-to-human language, the initial focus will be on teaching machines to recognize objects, navigate and make decisions.
Read more: DARPA wants robots that humans will trust