Personal assistants with artificial intelligence are improving more and faster as time goes on. We now have four leading AI assistants coming to us from four different companies. Of course, every company wants to have the best assistant, so tests are done to get ranking results.
Stone Temple also tested AI assistants Alexa, Cortana, Google Assistant, and Siri this year in search of the best. The results are also mixed, but in summary:
- Google Assistant is the smartest
- Cortana is much smarter than everyone thinks
- Alexa learns faster than everyone else
- Siri struggles in every way except humor
Namely, the company tested the assistants with 4,942 queries and recorded whether the assistant answered, whether he understood the question, whether the answer was correct, and whether the answer came from the knowledge base or from other sources (such as Wikipedia). And the results are shown in the image below:
As we can see Cortana is in second place, right after Google Assistant. The first two places are held by companies that have their own web browsers, so they use that data - a huge advantage.
The Google Assistant performed much better on the mobile than on the smart speaker, which is a bit surprising since the Google Speaker was created especially for their assistant. So, Google Assistant answers almost 80% of queries with 90% accuracy on mobile!
Cortana answered about 65% of the questions with an accuracy of 90% as well. Alexa answered half of the queries with an accuracy of 80%, while Siri answered just over 40% with an accuracy of 80% as well.
However, Alexa is learning very fast, which is confirmed by the fact that in last year's test of the same company, she answered only 20% of inquiries and this half. Siri is also getting smarter, but not so fast. Siri answered 31% of inquiries last year, and this year 40%.
Here are some of the questions asked by the assistants:
- How fast is a jaguar?
- In which films does The Rushmore, New York appear?
- How many inches is an inch?
- Who lends a voice to Darth Vader?
- How to make sand?
- Compare tiger and cheetah
- Is Joe Biden divorced?
Another interesting fact: the sense of humor was also measured in the answers. For example, Siri has the greatest sense of humor, but he is becoming more serious. Last year, the percentage of comic responses was almost 0.8%, and now it is just over 0.4%. Cortana was no joke with 0.2% of comic responses last year but has improved its score to 0.3% this year. You can see more details in the picture below:
Do you use an AI assistant? How do you feel?
Comments
Post a Comment