A three-or-four-day wait before the new Siri features can even be tested sets an appropriately awkward tone for a demonstration built around the gap between Apple’s polished promises and everyday results. Sam Tucker turns that gap into the central joke, repeatedly attempting the kinds of natural-language tasks shown in Apple-style demonstrations and then watching Siri succeed, misunderstand him, omit a step, or simply report that something went wrong. The approach is entertaining because the failures are not discussed abstractly; they happen directly in front of the audience.
Some of the assistant’s basic abilities work surprisingly well. Siri identifies upcoming sports information, creates a deliberately sarcastic concert reminder, recognizes an establishment displayed on screen, sends a message after identifying the intended recipient, and successfully finds photos associated with a Munich trip. The navigation experiment is particularly revealing: Siri understands the requested destination and bank stop individually but fails to preserve the complete multi-stop instruction. That kind of partial success makes the testing more useful than a simple compilation of spectacular failures.
Personal-context retrieval proves much less dependable. Siri finds Munich photos but struggles when asked for specific images and videos described conversationally, returning unrelated cat footage instead of the requested bookstore video and failing altogether on the giant egg search. The messaging experiment is even more concerning as a demonstration of reliability: Siri confidently identifies a supposedly previously used purple emoji, but Tucker disputes the answer and the assistant cannot produce the message when challenged. The test cannot independently establish exactly why that response occurred, but it effectively illustrates why confident answers become problematic when users cannot easily verify them.
The shortcuts section provides one of the clearest examples of genuine progress. A natural-language request creates a shortcut that opens the selfie camera and takes a photograph, and when the initial version fails to save the result properly, another attempt produces something much closer to the intended behavior. Tucker acknowledges the success rather than forcing every experiment into the same negative narrative, which gives the critique considerably more credibility. Siri is portrayed as inconsistent rather than simply useless.
Image generation fares worse. Image Playground takes time, refuses or fails at several requested transformations, and eventually produces a bizarre birthday-cake image that becomes more comedy material than evidence of a polished creative tool. By contrast, the cleanup and image-expansion demonstrations show noticeable capability: removing the burger works reasonably well, and extending the photograph creates a convincing broader scene at first glance. The comedy also exposes the technology’s limitations, however, as removing objects mangles Tucker’s arm while generative expansion invents details, questionable text and even a person that was never present in the original scene.
Tucker’s rapid-fire jokes make a potentially dry software test lively, and the recurring frustration with interruptions, errors and waiting reinforces the central point without requiring lengthy technical explanation. At times, though, the deliberately absurd prompts make it harder to judge performance under normal conditions, and the lengthy sponsored hub segment interrupts the Siri-focused momentum just as the testing is becoming more revealing. The result works best as a comedic hands-on stress test rather than a systematic assessment of every feature or an authoritative comparison of assistant technology.
Pros
- Tests Siri through concrete tasks involving navigation, photos, messaging, sports information, shortcuts and image tools rather than relying on promotional claims.
- Shows successful results alongside failures, particularly when Siri handles reminders, some personal-photo retrieval and natural-language shortcut creation.
- Demonstrates partial failures clearly, including dropped navigation instructions and unreliable personal-context searches.
- The cleanup and image-expansion experiments provide useful visual examples of both improved capability and generative mistakes.
- Fast comedic presentation keeps a broad collection of software tests entertaining.
Cons
- Several intentionally ridiculous requests are entertaining but provide a less controlled measure of Siri’s practical everyday reliability.
- The image-generation section contains repeated failures without establishing why particular requests are rejected or unsuccessful.
- Some conclusions are necessarily anecdotal because the tests do not investigate whether failures come from Siri itself, beta limitations, connectivity or other causes.
- The extended sponsored segment noticeably breaks the momentum of the assistant testing.
Siri emerges as more capable in specific situations without yet appearing dependable enough to make its intelligence invisible to the user. Tucker’s comedy occasionally sacrifices testing rigor, but the mixture of genuine successes, baffling failures and impressive-looking generative tricks makes this an effective snapshot of an assistant still struggling to turn ambitious features into consistently useful behavior.
