Apple Removes Siri Team Lead As Part of AI Strategy Shift (appleinsider.com)

The Apple executive who led the Siri team since 2012 has been removed as head of the project in a sweeping strategy shift favoring long-term research. Apple Insider reports: The Information reports Apple executive Bill Stasior is no longer in charge of Apple's virtual assistant team, though the executive is still employed at the company. Apple SVP of machine learning and AI strategy John Giannendrea reportedly made the decision in an attempt to shift the Siri program toward research rather than incremental updates. Giannandrea is anticipated to start a search for a new head of Siri, the report said.

Hired by former Apple executive Scott Forstall to run point on Siri, Stasior was previously attached to Amazon's A9 search arm. Stasior's removal as head of Siri comes at a critical point in the voice-enabled assistant's timeline. The first AI assistant to see wide adoption thanks to its inclusion in 2011's iPhone 4S, Siri's capabilities have fallen behind competing systems marketed by Amazon and Google. Apple is looking to Giannandrea to rectify the situation. Hired early last year, Giannandrea previously worked on artificial intelligence projects at Google. In December, he was promoted to SVP and put in charge of Apple's AI and Machine Learning programs, including Core ML and Siri.

Comments

Re: Maybe Siri should search for a new head of Sir

[Anonymous Coward]

Funny. What is it about executives... Can they not just tell this guy, hey we need to change strategies and do long term research? Is he not capable of adapting? Seems that at that level he should be able to take direction from upper management?

A Big Nothing Burger

[PKI Champion]

The person who led the team, Bill Stasior, obviously disagreed with the strategy change from above. He likely got into an argument where his pride got he better of him.

Re:A Big Nothing Burger

[Tablizer]

Rumor has it that Siri's AI is kind of out-dated and clunky. There may have been an internal debate about whether to incrementally retrofit the existing system, or deprecate it and start over. I suspect Stasior advocated for incremental, but seems the brass felt otherwise.

Re:A Big Nothing Burger

[mentil]

I suspect the bigger issue is that the Siri team hadn't produced any significant results over the past year, and wasn't expected to any time soon. Whatever gimmicks they were tooling around with didn't focus test well, didn't work, or were just obviously lame. Marketable improvements weren't on the horizon so they needed to go back to basic research for an indeterminate amount of time; that meant they might as well reassign all the people who were working on short-term improvements, into long-term research or other areas altogether.
Maybe they wanted something like Google Duplex, only via Siri. This could be made awesome for businesses. Instead of telling your secretary to call your dozen employees and e.g. tell them not to come in tomorrow, you can tell Siri to do it and all the calls happen at once, instantly, no secretary required.

Re:A Big Nothing Burger

[tlhIngan]

Or perhaps it's a decision of architecture - do you go big with a huge set of cloud-based servers that can answer any question under the sun, or do you try to figure out a way to make it mobile so Siri can attempt to do it all locally relying on local databases as well as Google and other internet databases?

Remember, Apple is trying to do more and more on-device and less and less in the cloud. Perhaps there's a difference of opinion - one wants to go big like Google and Amazon, while Apple wants to make Siri on-device. Yes, it won't answer questions as well as Google or Alexa, but if it completely works without talking to Apple's servers, that's a plus.

Re:A Big Nothing Burger

[LostMyAccount]

Is Siri's problem that it just doesn't work well -- doesn't understand requests or pulls wrong answers? Is it missing functionality that doesn't seem to require a lot more "AI", like voice-commands for accessing on-phone data?

For me I'd say it's both. I've never found it particularly good at interpreting requests, it doesn't seem able to integrate into my existing smartphone functionality well, and most of the time the answers boil down to returning search engine results.

Re:A Big Nothing Burger

[LostMyAccount]

There may be something to this -- Siri seems like it would be more useful if it could actually do things for me on my phone, which I think often means being able to process the data on my phone. This is probably a hard problem when you're relying on cloud processing but the relevant data is on the phone itself.

Re:Maybe Siri should search for a new head of Siri

[SeaFox]

Just sayin'.... if it's so fucking smart, it'll find a good candidate.

Someone didn't RTFS it looks like:

The first AI assistant to see wide adoption thanks to its inclusion in 2011's iPhone 4S, Siri's capabilities have fallen behind competing systems marketed by Amazon and Google. Apple is looking to Giannandrea to rectify the situation.

The whole reason they are looking for a new head is because, compared to Google and Amazon's Alexa, Siri isn't smart.

Good.

[wonkavader]

Siri is static, not progressing.

It has classic market-leader issues.

This doesn't make sense

[Rick Zeman]

in an attempt to shift the Siri program toward research rather than incremental updates.

You need to have both.

Re:Maybe Siri should search for a new head of Siri

[dgatwood]

The whole reason they are looking for a new head is because, compared to Google and Amazon's Alexa, Siri isn't smart.

Of course, IMO, the whole reason Siri isn't smart is that their privacy rules prevent the sorts of deep analysis that lead to it becoming smart. If the latter doesn't change, neither will the former, and no change of leadership is going to make any difference.

Sometimes, the best strategy is to do nothing at all, wait for a competitor to get good enough, and then license its services under terms you can live with, i.e. replace Siri with Assistant or Alexa, with whatever sort of anonymization you feel the need to layer on top of it (with the caveat that any anonymization will inherently make it inferior when answering questions based on things like personalized calendar events).

Either that or do absolutely nothing at all, and cede the field of services to companies that do it well, and focus on what you're good at,

Re:Tim Cook needs to be removed

[dgatwood]

Not just no, but h*** no.

Let me preface this by saying that I would not mind Forstall coming back as the head of the iOS UI design team. IMO, everything Apple has done to the user interface since Forstall's departure is approximately as abhorrent as you can get from a human interface perspective. The clean lines and flat simplicity result in a lack of depth cues that would otherwise make icons and buttons easily recognizable, which slows down user input significantly. And just about all the new features have been useless crap like Memojis (which are fine, but not as the *only* improvement). Bringing Forstall back and putting him in charge of the UI design team would be a positive improvement, because that is where his strengths lie. However, Apple should absolutely not put him in charge of iOS as a whole, and certainly not the company as a whole.

You see, Scott Forstall created the iOS culture that ran roughshod over the Mac culture within Apple. That new culture is basically responsible for everything that is wrong with the company today. Under his leadership, paranoia and internal power struggles grew massively, which really hurt the company by reducing the amount of feedback on products early in the design phase (when problems can be easily fixed). Of course, those problems didn't get fixed when Forstall left, because the people who took over didn't make any meaningful changes. Instead, the status quo continued, which is why things haven't gotten any better, and indeed, have continued to deteriorate.

IMO, the problems with Apple's post-iPhone culture run deep, and fixing them will be a big job for anybody who takes over the reins. I don't think for one minute that Forstall would be able to do that if brought back as a VP or CEO, much less be inclined to do so. Instead, I'd expect him to double down on all the things that are wrong, and generally make things worse until the company implodes.

So no, Apple doesn't need Forstall. It needs Wozniak. He's one of the few people who routinely calls Apple on its bad ideas, and I think he mostly shared S.J.'s vision for the company and its products. Whether he could be convinced or not is another question, but I can't think of anyone who would be better for the job.