Google has been developing its self-driving car for quite some time now. Several working prototypes have already been demonstrated (such as what is shown in the video footage below). The next step - as has been revealed recently by Google - is to upgrade the AI with human behavior.
The behavioral modification emerged as an important issue because it was found that the Google's robots tend to be overly cautious and obsessed with following rules. This frustrates human passengers in several completed test drives. For instance, there is the case of stopping. The current AI is programmed to assume the worst, and would often put on the brakes at slightest sign of danger, causing sudden stops from other motorists sharing the road.
The Google car also pauses frequently to collect its bearings, such as when it has to turn right or left. The current fleet of self-driving vehicles are "a little more cautious than they need to be," Chris Urmson, the head of Google self-driving car unit, said in a Wall Street Journal report. "We are trying to make them drive more humanistically."
By giving the Google cars human behavior, the AI will be able to cut corners, cheat in taking turns and limit the wide turns the vehicles are programmed to take, according to Jalopnik.
Human behavior is, of course, quite complex and it is not yet clear how Google will program its software or address possible traits, such as people's tendency to go beyond speed limit. Ironic that in Google's aim to design the perfect car for human-free driving, it must make the technology more human in order to be successful.
