In the video, Brownlee puts both devices side-by-side and activates them at the same time. He gives them a series of queries and commands, testing how each one responds. Brownlee starts with simpler tasks, like checking the weather, completing math equations, opening apps, and setting timers. He then moves into slightly more advanced queries, like asking what time the post office closes, Tesla's stock prices, and who the President of the United States is.
After figuring out who the president is, Brownlee asked both assistants how tall he is. Google Assistant was able to figure out who President Barack Obama was and how tall he is, while Siri had to resort to a Bing search. When asked how tall Obama was, Siri stumbled, searching for how tall the United States is. Brownlee soon asks how tall Obama is to both assistants more directly, and both answer the query with no problems.
Next, he once again tries to carry on a contextual conversation with the assistants. While Google Assistant could identify who won the Super Bowl and the team's current quarterback, Siri struggled to connect the two queries. However, when asked whether the Los Angeles Clippers won, Siri listed the most recent Clippers game, a preseason game, while Google Assistant listed a game from last season. Google Assistant also stumbled while trying to identify a song that was playing.
Brownlee then asks a series of contextual questions about the city of Kathmandu and Facebook CEO Mark Zuckerberg. Siri was able to keep up with Google Assistant, understanding context and linking together multiple queries, despite initially hearing "Zuckerberg" as "Sucker berg."
The video concludes with Brownlee trying to talk to both Siri and Google Assistant, asking how they are and whether they can make him a sandwich, tell him a joke, and more. Surprisingly, Google Assistant was more personable, joking around with Brownlee.
Brownlee found that Google Assistant felt more personable than Siri, understanding context better and able to joke around a little more. However, he felt that Siri provided him more information, using graphs and visual information. Siri, he said, presented him the info he needed and got out of the way, while Google Assistant was more talkative, reading the information out to him rather than relying on visual cues.
Apple has continually been improving Siri over the years, adding new functions and opening it up to third-party applications. In August, Eddy Cue, Craig Federighi and Phil Schiller explained that new machine learning techniques had cut Siri's error rate by half. Recent rumors have suggested Apple is aiming to improve Siri's functionality further for inclusion in an Amazon Echo-like smart home device. Apple has been hiring employees for its machine learning and AI divisions and acquired machine learning startups Turi and Tuplejump in recent months in an effort to improve Siri.