I’d like it to be known that I hated Siri before it was cool. I’ve written numerous articles on the subject. I’m an anti-Siri hipster. But when a bandwagon comes along, you better believe I’ll be sitting up front.
Earlier this week (as spotted by Daring Fireball‘s Jon Gruber) a Redditor named Guitar Scary started a thread about a one-off but deeply troubling experience with Siri. When asked the seemingly straightforward question “What month is it?” the hapless voice assistant responded, “Sorry, I don’t understand.” As Guitar Scary not unreasonably puts it, “Apple Intelligence this, Apple Intelligence that. Siri is still just awful.”
The commenters, predictably, tear Siri to pieces, joking about its inaccuracy (“Behold, Siri now uses advanced machine learning algorithms! So she’s smarter? She’s just stupid faster,” says one.) and offering their own similarly dreadful experiences. Those who repeated the experiment with the same wording got the same result, while slight variations in wording got answers that were wrong in different ways, such as “What month is it currently?” leading to “It is 2025.”
I tested this out for myself and had no more luck, although I too saw some interesting variations on the uselessness. In my first run “What month is it?”, “What’s the month?”, and “What month is it currently?” all resulted in flat responses of “Sorry, I don’t understand.” When I adjusted my wording to “What’s the month right now?” Siri changed things up slightly, asking “Do you want me to use ChatGPT to answer that?” (ChatGPT was able, eventually and via a different wording, to answer my question correctly.) When I tried yet another tack, asking “Is it January?” Siri showed me calendar entries for the previous January rather than answering the question with a flat, “No.”
You can type to Siri now but that hasn’t made it any smarter.
Foundry
But the badness varies from attempt to attempt as well as from user to user. On a second run, things were somehow worse. “What month is it?” and “What month is it currently?” were the same, but “What’s the month right now?” no longer offered to bring in ChatGPT. And “Is it January?”, unforgivably, produced the answer “It’s Thursday, 1 January 2026.” (For the benefit of anyone reading this article in the future, it’s currently March 20, 2025.) A colleague also got that last answer on some attempts on both iPhone and Mac.
Presumably, Siri thought I was asking when the next January begins, but it’s a fairly basic question it should be able to get right after 15 years. And I have no explanation as to why it can reliably tell you the month… provided you ask instead for the day or year first. In either case, it confidently provides the day of the week, the date, the month, and the year. All correct, too. But this doesn’t reassure me. I imagine Apple will respond to this bad publicity by making sure Siri can tell you the month… but the fact that after all these years it still needs specific programming indicates far more worrying systemic failings. Siri needs to be accurate, consistent, and intuitive. At the moment it is none of these things.
Other questions it didn’t answer correctly: Who won the World Series last year? (The Rangers won the World Series 4-1.) When do the NBA Finals start? (The Celtics won the NBA Finals 4-1 against the Mavericks.) Who was the president last year? (The incumbent president is Joe Biden, who assumed office on January 20, 2021.) Who is the NFL MVP? (I can use ChatGPT to answer that.)
Siri has been bad for years, and there can be few optimists remaining who seriously expected it to suddenly get better. What is surprising is that it’s actually getting worse. And people are noticing. This needs to change, and soon.
How does Android compare?
Before we get out the pitchforks and flaming torches, we should run a comparison with the tech currently offered by other companies. Maybe “What month is it?” is just a super-hard question for AI/voice assistants to handle.
Yeah, no. My pal and colleague Anyron Copeman from TechAdvisor asked that very question to three Android phones equipped with Gemini, Google’s equivalent of Siri. And got the following answers:
Anyron Copeman / Foundry
See, Siri? It’s not that hard.
So, what’s the answer?
Where next for Siri? As with so many tech matters right now, the answer is apparently to throw more AI at the problem. Apple expects this to make Siri more accurate, not to mention more conversationally adept, which should help when dealing with questions like this for which it does have the data, if you only knew how to ask for it.
When Apple Intelligence was first announced, one of the most intriguing elements was a raft of improvements to Siri, including the ability to leverage contextual data (such as other information on the screen, facts it knows about the user, and things previously mentioned in the conversation) to handle queries more effectively. It turns out Apple jumped the gun with this feature, which has been delayed until at least 2026 (a setback that the company reportedly regards as “ugly and embarrassing”) and still hasn’t been demonstrated to anyone outside the company.
With the new Siri looking further away than ever, it’s tempting to wonder if Apple even knows how to fix the problem. It had a huge head start in the voice assistant market but has squandered its lead and any goodwill Siri once held as other assistants surpass it in every department and Siri itself seems to get worse with each passing year. Yet the success of other assistants shows that the tech is there, and Apple is hardly short of talented employees, or the cash to buy its way back into a strong position if all else fails.
But one thing is certain. If Apple doesn’t already regard fixing Siri as a top priority, it should do so sooner rather than later.
