Hacker News Reader

Comment by late25 | Hacker News Reader

>It’s helmed by John Giannandrea, Apple’s head of AI, who was hired in 2018 to help improve Siri.

I’m sure he’s a smart man, but considering where Siri is, and the fact he’s been there for 5 years, I’m not holding my breath.

View on HN

At a big company, unless you have some serious lubrication, the first 12-18 months are spent landing. Usually you have to land your big changes in a planning cycle which are usually 6-12 months. This often means, for your first ~2 years, you're not able to do much.

That, plus the lost years of productivity during COVID -- I'd say it's okay to forgive for not turning the ship.

We are talking about 5 years, which is longer than the average tenure at most tech companies.

I also don't buy any Covid-related excuse with this one. If anything, I personally saw at least a couple companies make good use of the "never let a good crisis go to waste" mantra, using the general chaos of the early months of Covid to make some long-standing necessary-yet-painful changes - that is, changes that pretty much everyone knew were the way forward in the long term, but had risk of hits to revenue or much higher expenses in the short term. Using the pandemic as an excuse to go all-in on AI/Siri (e.g. "People are spending more time at home/on screens and we need a better voice interface") would have been the perfect approach IMO.

You're expecting incremental improvements.

Apple sometimes does incremental improvements, and sometimes does an entirely new release.

Five years to improve? Sure, probably could've done a lot. Five years to start completely new? Not enough time to make it land with a bang compared to the existing Siri.

Yeah when it's a new product it makes sense getting it right before releasing. But leaving their customers with crap for 5 years does not make ton of sense.

I guess there's precedent in that they did the same thing with the butterfly keyboard

There was no lost productivity due to COVID in software other than than some short term pain. Doesn’t make up 5 years of nothing

>but considering where Siri is

Siri is honestly so bad, I don't use it. Every time I try to schedule a meeting it tries to incorporate it in the calendar and looks for a contact. If I say "put meet with Mike on my calendar for 2 PM on Wednesday" it'll come back with "I can't find Mike on your contacts." Unless that's changed. Then when you ask it a question its 50/50 if it answers or just gives you useless "this is what I found on the web."

I've started using the ChatGPT app with Siri and it feels like how it should. Only problem is it can't do the scheduling or other useful things.

The tts in the ChatGPT app is also much better than Siri

> I’m sure he’s a smart man, but considering where Siri is, and the fact he’s been there for 5 years, I’m not holding my breath.

This is the standard Apple playbook. Five years is nothing for a new Apple product or foundational technology. The M1 was based on key ARM advances that Apple started building 10 years before it was introduced. Apple worked on the project that became the Vision Pro for 16 years.

> Apple worked on the project that became the Vision Pro for 16 years

So during Steve Jobs tenure? Can I conclude - it got eliminated from his biography on secrecy grounds?

> Can I conclude - it got eliminated from his biography on secrecy grounds?

It'd be more accurate to say that biographers would not have heard anything (from Steve or any other in-the-know Apple employees) about any projects that hadn't already been unveiled by Apple.

Additionally, it seems likely that biographers would've been required to not ask about possible future projects hinted at by public resources like patents: https://www.patentlyapple.com/2023/08/apple-won-a-patent-tod...

If they'd hire the right person, they could have Siri in the top 3 of assistants within the timespan of a perhaps few months. It's not rocket science anymore and in fact most of the underlying tech is already open source.

Products at Apple have to meet a high bar for all of: Privacy Compliance,Public Relations+Marketing alignment,World Class R&D,Focus Testing,UAT and Deployment on billions of devices. I don't see how one person could make that big of an impact. And in a few months? Somehow I don't think you thought this statement through very well.

I agree, but we’re still talking about five years for a huge company with colossal cash reserves. Surely one person won’t cut it to do all the work, but I’d wager that the person overseeing the Siri department isn’t doing their job very well. Sounds more like there are managerial/political issues rather than technical ones.

"I could build Facebook in a weekend."

I dunno, but someone like Karpathy could build a better Siri, with the help of a bunch of engineers.

The big problem with the competitors is that Amazon and Google for the most part subsidize its usage . Microsoft and Google took a look at what OpenAI and jumped the gun on both projects. Siri has been ok the back burner for awhile but launching a product too early is what Apple tries to play off of but the neglect of Siri is still a huge disadvantage .

Indeed. Same thoughts. I think you need business & product smarts here not raw engineering smarts.

Eh, who knows really? Having something in the lab that works and getting a product team to understand how to make it useful to customers aren’t the same problem.

In some sense I’ve always thought Apple is focused so much on reliability and deterministic approaches that wrapping their heads around probabilistic outcomes is harder than for other companies.

Yeah, he is a Google hire. But I am confused as I (mis)remember he left Apple on WFH issue.

That was just an engineering manager that all the press misrepresented as “head of AI” to make a more juicy story.

Reply on news.ycombinator.com