While the Danish law prescribes that it under no circumstances is legal for drivers in vehicles to use any handheld mobile phones, the need for using a mobile phones while driving, is still there. While there are many different ways of which people choose to interact with their phones while driving, many of these solutions are problematic in regards to the law. Intelligent Personal Assistants (IPAs) have received more attention than ever, both within the field of HCI, and people's everyday lives. Apple, claim that their IPA, Siri, can help people complete tasks easier and faster, and the literature within the field suggests, that IPAs, especially make sense to use while driving, because the hands of the driver are otherwise engaged.
In order to investigate this matter, we set up an experiment with the means of evaluating the usability of Siri while driving. This was done by comparing different conditions: use of Siri while driving, manually interaction with an iPhone in a car, after having pulled over, use of Siri in lab and lastly manually interacting with an iPhone in lab. The methods used were a questionnaire, a qualitative interview and a usability test. To measure how the participants perceived the usability, we deployed eye-tracking and video recording as techniques for data collection. Results show that the use of Siri while driving has a negative effect on the usability. The combination of interacting with Siri while driving requires, much cognitive effort and visual attention from the users, which at times even resulted in frustrations. A frequent issue the participants had with Siri, was often related to the voice recognition not being able to detect what they said. This was especially an issue for female participants when driving, but not as prominent in the lab condition. While completing tasks manually in car appeared to be the better option in terms of usability and especially safety, this option is not always available, which can be interpreted as a need for an alternative solution. For that reason we have proposed a list of suggestions that may lead to improvements for the usability of Siri while driving. Some of the conclusions from that list, include that Siri needs to be able to take the context of use into consideration. As an example, answers from Siri should always be read aloud when driving. Since multitasking is demanding and present when using Siri while driving, we also suggest that the time window in which users have to formulate their requests should be extended, giving them more time to think and formulate requests. In our study, we found that there are pros and cons associated with either a manual way of completing tasks or completing them with Siri. For future work, it would be interesting to look deeper into how frequent Siri users are able to perform and whether they would perceive the usability of Siri while driving better, because of their experience with it.