Slack CEO denigrates Apple's Siri in announcing work on own team-serving virtual assistant
Slack Technologies founder and CEO Stewart Butterfield on Monday revealed plans to invest in artificial intelligence with an eye on virtual assistant tech for work groups, while calling Apple's solution "nearly useless."
In an interview with The Wall Street Journal, Butterfield said future iterations of Slack's existing Slackbot "robot" will incorporate natural language processing and machine learning algorithms to help deal with tedious work tasks. The improvements, he said, will result in a system more akin to the advanced AI portrayed by actress Scarlett Johansson in Spike Jonze's movie "Her" than competing products from Google and Apple.
"Apple spent billions of dollars on Siri and worked on it for a very long time with hundreds of engineers and a huge dataset of voices - and it's f-ing idiotic. Siri is nearly useless," Butterfield said.
Instead of sinking time and money into product R&D, Slack is looking to partner with firms focusing on AI technology. To that end, the company recently hired its first data scientist and plans to invest in the artificial intelligence space.
With a next-gen Slackbot, Slack is looking to create a virtual assistant capable of tapping into a customer's institutional infrastructure, while connecting simultaneously with everyone on a team to automate certain tasks like scheduling meetings. The goal is to remove time consuming, banal tasks from an employee's plate, something Butterfield believes could boost productivity by 20 to 30 percent.
Current iterations of the programmable query/response system are largely rooted in Slack's direct messaging application, but the bot also surfaces searches and other productivity-minded applications. Future versions might be able to scan a Slack team's messages, recognize when an employee is having trouble with a particular task and automatically schedule an appointment with higher ups to keep the project on track, for example.
Slack is just one of many tech companies conducting research into natural language processing and machine learning. Apple's Siri is one of the best known virtual assistants on the market, but competing services from Google, Microsoft and Amazon are finding success with their respective platforms. Facebook is also researching its own personal assistant called MoneyPenny, a natural language and machine learning initiative that studies human interactions across the social network's vast user base.
In an interview with The Wall Street Journal, Butterfield said future iterations of Slack's existing Slackbot "robot" will incorporate natural language processing and machine learning algorithms to help deal with tedious work tasks. The improvements, he said, will result in a system more akin to the advanced AI portrayed by actress Scarlett Johansson in Spike Jonze's movie "Her" than competing products from Google and Apple.
"Apple spent billions of dollars on Siri and worked on it for a very long time with hundreds of engineers and a huge dataset of voices - and it's f-ing idiotic. Siri is nearly useless," Butterfield said.
Instead of sinking time and money into product R&D, Slack is looking to partner with firms focusing on AI technology. To that end, the company recently hired its first data scientist and plans to invest in the artificial intelligence space.
With a next-gen Slackbot, Slack is looking to create a virtual assistant capable of tapping into a customer's institutional infrastructure, while connecting simultaneously with everyone on a team to automate certain tasks like scheduling meetings. The goal is to remove time consuming, banal tasks from an employee's plate, something Butterfield believes could boost productivity by 20 to 30 percent.
Current iterations of the programmable query/response system are largely rooted in Slack's direct messaging application, but the bot also surfaces searches and other productivity-minded applications. Future versions might be able to scan a Slack team's messages, recognize when an employee is having trouble with a particular task and automatically schedule an appointment with higher ups to keep the project on track, for example.
Slack is just one of many tech companies conducting research into natural language processing and machine learning. Apple's Siri is one of the best known virtual assistants on the market, but competing services from Google, Microsoft and Amazon are finding success with their respective platforms. Facebook is also researching its own personal assistant called MoneyPenny, a natural language and machine learning initiative that studies human interactions across the social network's vast user base.
Comments
Siri could be better. A LOT better.
I also think that he, like us, does not realize what is happening behind the scenes. Apple I suspect is well aware of the huge limitations with public Siri and they are intentional.
For one they want what it does do to work, not just pretend it can do everything and suck at it,; for two Apple is improving what is does do steadily; so that for three they will release a much more capable version of Siri based on all the research they are doing on what Siri is dealing with now.
We haven't yet had a real Siri 2.0 moment, everything has just been point releases.
I also personally think that using song/artist/album names and dictation as the root points of getting better at a)comprehension and b)building context algorithms based on how people speak out sentences is critical for next level assistants.
And I would put money on it that Apple is using all our interactions with Siri for this kind of 2.0 R&D
I also, finally think that Apple has been putting energy into building out a global back end that can support said system for 100's of millions of users. AND working on custom silicon for local processing abilities and that the development of these two key components plays a part in Apple appearing to drag its feet with regards to a Siri 2.0
This makes so much sense. Most of the people pressing me to use Slack are kind of dickish.
Siri gets better everytime I use it. We were using it at work today to play music from both popular and obscure groups at random. She found and played everything we could think of to throw her. We were shocked. Siri does things like prompt you to leave early for an appointment if you put a location and traffic gets backed up. All without being too annoying.
When I see cocky companies like this I hope Apple buys them(If they're not bluffing). They'll shut up real quick and Apple takes the work, and they work for their dream company and make money.
Also, Watson?
1). WTF is Slack?
2). They just hired their first data scientist. LOL
3). Why is AI reporting this?
I really don't have a rebuttal to this. Siri is really, really good at telling me what time it is. Anything else stumps the shit out of her.
I use Siri pretty often and it generally works very well. I do wish, however, it could stay localized on the phone when I don't need it to connect to the internet. For example, dictating a text or setting a reminder or playing a song that's already on my phone.
But I've noticed a trend. Tech company CEO, "Hmm, what can I do to get tons of free press? I know, call an Apple product $#!+"
It's embarrassing
Slack Technologies founder and CEO Stewart Butterfield ...
Who?
I do not use Siri at all (except when somebody asks if 'this has Siri'), just the dictation feature once in a while. Speech recognition in German and English works fine, can't say much more. We all know that it e.g. does not have all the features Google Now offers, but many of those are not there because of 'philosophical' differences when it comes to handling private data. As I won't give Google (or Slack) any of my data, that is cleared.
What irks me about Butterfinger's rant is that the whole comparison he entertains does not make any sense. Siri has to deal with arbitrary questions, mostly without knowing any context. A line of business application, and especially something as trivial as task tracking and meeting scheduling, has plenty of context available. The list of 'players' is known, the project/task data is known including progress and overdue status, all calendars are accessible etc ad inf. This is not more than modelling some swim-lanes and making sure the software finds the right entry point. I do not see how this is harder, or more advanced. (And, having worked on such applications myself in the past, I might add that the solutions are fairly trivial 99% of the time, as long as you can get somebody 'upstairs' to decide on a procedure in the first place.)
BTW, this guy is pretty happy with Siri:
Siri Saves Tennessee Teen After He Gets Pinned by Truck - NBCNews.com
Maybe today wasn't the best day to call Siri "nearly useless"?
Not that AI would cover a positive story about Apple...
I have found Siri to be pretty good and gets better all the time. iOS 9 will bring even more improvements.
But I've noticed a trend. Tech company CEO, "Hmm, what can I do to get tons of free press? I know, call an Apple product $#!+"
Yes, trash talking Apple brings immediate press, investment... More so than their actual shitty product will in 2 years when its released that's for sure...
Again, a la so many such software and networking innovations (too numerous to list), Apple initially gets it going out there better than any one else, but then lets others catch up or even overtake them.
I do not use Siri at all (except when somebody asks if 'this has Siri'), just the dictation feature once in a while. Speech recognition in German and English works fine, can't say much more. We all know that it e.g. does not have all the features Google Now offers, but many of those are not there because of 'philosophical' differences when it comes to handling private data. As I won't give Google (or Slack) any of my data, that is cleared.
What irks me about Butterfinger's rant is that the whole comparison he entertains does not make any sense. Siri has to deal with arbitrary questions, mostly without knowing any context. A line of business application, and especially something as trivial as task tracking and meeting scheduling, has plenty of context available. The list of 'players' is known, the project/task data is known including progress and overdue status, all calendars are accessible etc ad inf. This is not more than modelling some swim-lanes and making sure the software finds the right entry point. I do not see how this is harder, or more advanced. (And, having worked on such applications myself in the past, I might add that the solutions are fairly trivial 99% of the time, as long as you can get somebody 'upstairs' to decide on a procedure in the first place.)
Yes, context/task simplifies things massively; his whole rant means he's trash talking Apple for media attention.
Its like saying Deep Blue (hyper specialized context and vocabulary (chess moves)) would beat "Watson" at chess! Probably!! ;-).
But, I'm guessing Watson is still better at teaching someone how to play chess :-).