Loup Ventures has released its annual IK tests of intelligent intelligent speakers, revealing that four large personal assistants AI performed when asked for 800 questions. Compared to the previous year, three products achieved the highest gain in IK points and managed to answer a significantly higher percentage of questions compared to 2017. However, Google Assistant still mastered them.
The study was quite simple, including 800 questions presented by the AU with two categories for the results: is the question properly understood, and if yes, is the right answer offered? Google came first in the first place, and Siri was dignified by another.
Google Assistant understood 100% of the questions asked and could give the correct answers for 87.9% of them. Apple's Siri is second with 99.6% of comprehensible queries and 74.6% of correct answers.
Amazon's wild popular Alec scored in the third slot of the study, realizing a solid 99% of the questions and only correct answers for them 72.5%. Finally, Microsoft's Cortana was the last in the category "correctly answered" to 63.4%, but managed to transfer Alec to the third slot in a "reasonable query" with a 99.4% rate.
Regarding the exactly answered questions, all four products made an improvement in 2018. Aleka increased 64%, Google Assistant with 81%, Cortana with 56%, and Siri with 52%.
Breaking down this year's results, Loup Ventures has discovered that almost all the wrongly asked questions have a real noun and that it is usually a local restaurant or city. However, it has been observed that natural language processing and voice recognition have improved among all four assistants.