The first step towards a (distant) future where these voice assistants might be able to provide valuable support in health-related scenarios is understanding exactly what comes out of the user's mouth. Unfortunately, both Siri and Alexa are constantly struggling to recognize commonly dispensed medication names in the US, as evidenced by new Klick Health research published in Nature Digital Medicine.
Said research awarded Apple's AI an overall accuracy score of 58.5 percent for medication brand-names, as well as 51.2 percent in terms of correctly identifying generic names of drugs. While that essentially means Siri was right most of the time, the two numbers are hardly impressive for a nearly 8 year-old piece of software that should be able to quickly scan the web for answers to all your questions.
The same goes for Amazon's Alexa, which scored even lower accuracy rates of 54.6 and 45.5 percent for recognizing medication brand-names and generics respectively. Incredibly enough, Google Assistant completely obliterated its competition, correctly identifying generic drug names 84.3 percent of the time while boasting a 91.8 percent accuracy score for brand-names. That's still not perfect, but it's yet another area where the search giant's 2016-released digital assistant shines.
That shouldn't come as much of a surprise, though, as Google crushed Apple and Amazon before in general reports judging the overall "IQ" of these assistants based on questions covering several large categories of information, from commerce to navigation to command skills. Google Assistant was also found last year more proficient at understanding accents, which actually comes up in the new research paper as well.