Believe it or not, Apple’s Siri is actually quite good at lots of things, recent data shows

“It’s hard to be Apple’s virtual assistant because lots of people complain about what you do, even when you’re actually quite good at lots of things, recent data shows,” Jonny Evans writes for Computerworld.

“Virtual assistants are destined to do much more than send memos, capture shopping lists or tell cheesy jokes — they will become one of the primary ways we interact with the ambient AI that will surround every part of our lives in the not-too-distant future,” Evans writes. “With so much at stake, public image means a whole lot, and that’s why I found it so interesting to take a look at a recent virtual assistant customer satisfaction study from CBT Nuggets.”

“They asked 500 people to rate their experience of using Siri, Cortana, Google Now, Google Home and Alexa for various tasks and in different situations,” Evans writes. “Overall, despite all the mockery Apple’s virtual assistant has to endure, Siri delivered the highest percentage of perfectly completed device responses in both quiet and noisy environments. It gets it right 70 percent of the time overall.”

Read more in the full article here.

MacDailyNews Take: With 65% overall satisfaction rate, Siri has plenty of headroom for improvement. Google Now and Cortana have even more room, with just 59% and 46% overall satisfaction respectively.

With machine learning, we expect Siri to show even more improvement than it has recently. Those who use Siri regularly likely agree that Siri has shown noticeable improvement in the last few months.

29 Comments

  1. “It gets it right 70 percent of the time overall.”

    I read something years ago by a computer user who said, in effect, that if his typist was correct only 90% of the time he’d fire them.

    1. True but that typist took time to go from under 50% to anything like 90+%. It’s just that computer user didn’t do the work to get her there just exploited the results.

    1. Ask Siri to set an alarm for ten-to-two. Good luck 😉

      We were watching Gogglebox the other night and the sponsored advert was an Amazon Alexa avert and this is how it went down:

      Paid shill: “Alexa! I am going to heat a pizza in the oven.”

      Alexa: *crickets*

      Paid shill: “Alexa, please set an alarm for 25 minutes.”

      Alexa: “Okay, alarm set for 25 minutes.”

      @end

      Wuaaa? The pizza bit had no contextual point at all but they put it in anyway?

      (P.S. who the f*ck is the responsible for the maintenance of MDN’s website? It SUCKS)

        1. To be honest – although I did temp the poster to do it – I am lost at your reply?

          The article in large said that Siri works 70% of the time.

          I pointed out a command that didn’t ever work for me, and you pointed out that it did for you. Just us 2 people are a tiny subset of data points. Shall we conclude that it never works (for me), that it works 100% of the time (it worked for you) or based on our very tiny study that now Siri works 50% of the time?

          The article at large points out that a very big study concluded that Siri works 70% of the time…

          Again what was your point?

        2. You seem to think any worthwhile test would only include asking that specific question and indeed asking it of only two people for your own answer to make any sense. One instead needs to ask a lot of people a lot of varying questions and get an overall score and that’s the point he was surely making. So I in turn am lost at your reply as it bears no resemblance to a real world of indeed a lab test, just a tiny snippet used to predict a far more complex set of results. You would not conclude the winner of the Prem League based on the results of the first weekend’s matches.

        3. You are preaching to the choir. You contradicted yourself, by pointing out to me, what I pointed out the other poster. And to be fair 85% of my initial post was actually ripping it of Amazon’s Alexa (based on true events (facts)).

          Get of me, and don’t make make “the beautiful game” based analogies when speaking to me about tech.

        4. My Alarm set to 1.50am first time as well, therefore Siri was successful 66.67% of the time. If only you hadn’t posted in the first place Ron Burgandy then Siri would be showing a 100% success rate. Except if you hadn’t posted I’d have never set my alarm to 1.50am. Will Siri understand the command Siri “Alarm STFU” at 1.50 on Tuesday morning? That’s the real question.

        5. If I just say “Set an alarm for ten to two” it sets for the next upcoming instance of ten to two, which for me was Tuesday, and I guess Siri is smart enough to assume an alarm at 1:50 am doesn’t make sense for me. If I say “Set an alarm for ten to two on Thursday” Siri puts that into Reminders for me and again assumes 1:50 pm. Although I can say “Set an alarm for ten to two AM” and Siri gets that right.

          Siri is far from perfect, but Siri is not nearly as bad as a lot of the doom and gloom articles I see.

    2. Well for me of late yes, the difference to a year ago is amazing, it’s understood and faultlessly responded to my commands since I started testing her again this past month or so. If nothing else it’s great to simply ask her to play the music I want on my Mac rather than break from what I’m doing and similarly answering a question when my partner or I hear something on the radio that posed a question, one of which recently was totally obscure and I was amazed siri immediately understood and gave the correct info. Depends on what sort of mind you have as to the usefulness. Been equally impressive on my elderly iPad understanding my questions on one occasion, despite me stumbling over it. Yes at last I am impressed and start to seriously appreciate the potential power in the future once that accuracy approaches 85 to 90%. There is a tipping point for becoming seriously useful for many of us, and it’s getting closer by the months let alone years. Finally they seem to be pulling their finger out. About time mind.

  2. I have both Siri and Amazon’s Echo. I find that Siri is more than useful and I would really miss it if it were gone.

    I find Siri to be more useful for automation and personal management while Alexa is more useful for BS like, “Alexa, News” or “Alexa NPR 1.” I enjoy Alex’s apps, aka skills.

    The mix of Siri and Applescript is extremely powerful also.

  3. Would be nice to see what the questions/tasks were even if the sample size was only 500. Or did they mean 500 users of each VA? The source article was not too clear on that.

  4. Most of the industry regards Siri as being the least useful voice assistant although I’m not sure who in the industry set the standards. All I know is that Apple let Siri stagnate and allowed Amazon to take over that entire market. It’s possible even Apple didn’t think Siri was worth the time and effort to improve. I’ll never understand why Siri was only put on the iPhone and not any of the desktops or laptops. What was Apple’s purpose in limiting Siri to only one device?

    Apple could have had many thousands of more devices using Siri but instead, let Amazon and Alexa become the market leader and get all the praise and glory. Siri went from first to last without so much as a whimper.

    1. Maybe the plan was to make Siri’s personalization data secure on the Secure Enclave. That would have excluded anything but an iPhone to that roadmap for a good number of years.

  5. Two areas Siri will need to catch up with Google Assistant is in ‘context’ questioning longer than 2 questions and being able to use Siri by typing for those times you don’t want to have people overhear.

    Context questioning is basically where you can ask a series of questions that take context from the questions asked it earlier in the ‘chain’. For example, “Who is the President of the United States?” followed in sequence by “Who is his wife?”, “How many children do they have?”, “What are their names?”, “Do they have a pet?”. Siri tends to fail after the 2nd question.

  6. I’ve limited my Siri commands to a very limited subset that I know should work (alarms, timers, workouts). I still have the problem (on my Apple Watch) with the message to “hold on” and quite often “Siri is unavailable”. I wonder if this has something to do with my VPN that I installed earlier this year, or if my gen 1 Apple Watch and iPhone 6+ are just showing their age. Shitty.

Reader Feedback

This site uses Akismet to reduce spam. Learn how your comment data is processed.