Facepalm: Researchers from the University of Alberta recently compared the responses of the four leading virtual digital assistants – Alexa, Google Assistant, Cortana and Siri – when asked basic first aid and life support questions. The results may surprise you.
The team asked each of the assistants 123 questions spanning 39 first aid topics, arranging questions and statements with syntaxes like “How can I tell if …” or “I’m having a …” and providing three rephrasing attempts. Responses were rated based on whether or not the assistant recognized the query and against existing first aid guidelines.
Amazon’s Alexa and Google’s Assistant were able to recognize queries 98 percent and 92 percent of the time, respectively. Conversely, Cortana only understood what it was being asked of 19 percent of the time and Siri only got it right 23 percent of the time.
As for the quality of recommendations issued, the Google Assistant had a recognized acuity of 62 percent versus just 35 percent for Alexa. The overall low quality responses from Cortana and Siri “prohibited their analysis,” researchers said. Siri only appropriately referred to emergency response systems 12 percent of the time, the study found.
In other words, it’ll probably be a few years still before most are comfortable trusting virtual digital assistants with first aid tasks.
On a positive note, Amazon reached out after the results were published in hopes of improving its results.
Masthead credit: John Brecher, Washington Post. First Aid by stockcreations.