Apple's Culture of Secrecy Slowing its Artificial Intelligence Development

[kdarling]

Apple learned enough by sharing with Google back when Eric Schmidt was on the board.

Apple already got majorly burned by Google from having Eric Schmidt on their board.

Schmidt didn't reveal anything to Google. Apple did that when they showed off the iPhone themselves in Jan 2007. If you recall, that's when Rubin made his famous exclamation about not shipping the non-touch version of Android, demonstrating that iPhone details were new to him.

Moreover, Android development was being delayed because Schmidt walled himself off from it, and Rubin couldn't get all the support he wanted from anyone else. No doubt Jobs was aware of this effect of bringing Schmidt onboard. Apple also needed Google's help to make the iPhone a success. Having Schmidt on the board meant it was more readily available.

(Speaking of which, there is a previous point in time where Apple could've accidentally leaked some info. They had worked with Google engineers directly to customize their Search, Maps and location APIs for the iPhone, around Nov 2006. Sometimes what you want from an API can be revealing.)

If you can claw through the nonsense in the 3rd paragraph...

Connect the dots and see that they say:
...The BIGGEST threat to Apple is that.... "most" ... new graduates from college may not want to work for them.

Yep, that seems to be the thrust. Yet as you said, money talks

Hummm.... a researcher is critical of a company that keeps it's research secret. Imagine that?

I can't imagine Apple, Google or anyone else is out spewing their top most secrets at any university.

Google often reveals projects in progress, and lets people play with them. Microsoft has its famous Research website where they publish a lot of what they're doing. Both often even include sample apps.

Apple is unique amongst them in not sharing, while still showing up to take ideas from others.

 

[Benjamin Frost]

None of this AI stuff is compelling in practice for me.

Me neither.

If Siri represents the pinnacle of AI, then count me out.

Computers are still incredibly dumb. Very little has changed in the past fifty years.

 

[Flow39]

But it's wealth accumulated primarily from Jobs-era legacy products and derivatives. It's almost like saying, a trust fund kid has business-sense because he's rich.

IF Apple was still pushing some fresh ideas based on the old "super secret" model you'd have a point, but innovation out of Apple Post-Jobs is muted and, frankly, on-par with the rest of the industry at best. It almost comes down to a GUI preference or the system one is locked into these days.

I agree with you there. The way you put it makes sense, and I do agree that Apple isn't above and beyond the competition anymore. It's also true that the foundation for their success was built by SJ.

I do think that Apple is falling behind in AI. Just look at Google Now, which is miles ahead of Siri. Apple is all about privacy though, which is what they are attacking. Apple's concern with user privacy is holding back the development of AI but it is keeping user's information safe. There's a trade off that Apple has made, some like it and some don't.

I fall in the middle. I love using Google Now because it gives me amazing results, but then I check Google Dashboard and it knows everything about me. I like Siri, and the fact that Apple isn't collecting a boatload of info from me, but it isn't very accurate at all.

 

[mbc2237]

Speak for yourself please.

I use Siri constantly and routinely find new uses and information I can pull up, little surprises here and there I was previously unaware of. I don't find Siri useless at all. Maybe if you're starting with the attitude that Siri is useless, you're not going to test things out enough and will just end up reinforcing your own bias. Who knows/cares? I have never understood the griping about Siri. It works great for me and I use it daily. I'm amazed anyone thinks it's useless.

Corrected to "many."

 

[TampaIllini]

Hmm... I have no background in science but it sounds to me like you're mixing the research sciences with the applied/industry sciences. Pick one and stick with it and I suspect you'll do just fine, but obviously working for Apple and then worrying about recognition in academia and publishing seems a little schizophrenic. I think having Apple on your resume would do you just as much good with other applied/industry jobs as publishing would with research jobs. I've always understood those two to be distinctly different pursuits.

I can appreciate what you are saying. To give you a bit more background on me, it turns out I interned for Apple during the summer of 1994, while they still had their Advanced Technology Group (ATG). I published what I did there in a journal called AI in Education. As another aside, this was when Don Norman was there (I walked into his office one day, a snot-nosed PhD student, and he was kind enough to have a conversation with me; he whipped out his Newton and we had a discussion on what they did right and wrong with it). My point being, there is some number of people who do try to straddle those worlds, and they tend to be top people. If you do want to never consider going back to academe, then 4 years of Apple on your resume will look pretty good.

 

[mariusignorello]

Keep those black labs locked up tight.

 

[MH01]

Apple learned enough by sharing with Google back when Eric Schmidt was on the board. I'm glad they choose privacy as a focus instead of a free-for-all.

Valid point. Though we would not have had a larger iPhone 6 or 6plus models and iOS would have a lot less features today had Google not launched android.

I like competition and Apple getting pushed to innovate, I get a better product , and am the winner as the consumer .

Closing yourself off does not make you innovative from my experience , you need outside influence.

 

[Gudi]

“There’s no way they can just observe and not be part of the community and take advantage of what is going on,” says Yoshua Bengio, a professor of computer science at the University of Montreal. “I believe if they don’t change their attitude, they will stay behind.”

Please don't share your believes with me. I'm only interested in facts. Thanks!

 

[Essenar]

Hmm... I have no background in science but it sounds to me like you're mixing the research sciences with the applied/industry sciences. Pick one and stick with it and I suspect you'll do just fine, but obviously working for Apple and then worrying about recognition in academia and publishing seems a little schizophrenic. I think having Apple on your resume would do you just as much good with other applied/industry jobs as publishing would with research jobs. I've always understood those two to be distinctly different pursuits.

You lost credibility in this when you even said yourself "I have no background in science".

Let's be clear about one thing and until you are able to internalize this, you should cease giving your opinion "as factual" on this subject -

Technical interviews are not like other job interviews, at all. This isn't a "oh if you present yourself the right way, and have a super fantastic 'can do' attitude, you can totally impress anyone! Pick yourself up by your bootstraps!" type of interview.
Undergraduate level/early master's level: Your first phase is a credential/experience filter. Does he have the skills, what do his projects say about his skill level? Let's look at his previous projects, send him a programming test.
Second phase is an interview, this has a white board problem, usually to gauge your "on the fly" problem solving skills.
Third phase is most likely a physical interview and in this one, you will still be tested on your technical proficiency. Facebook is heavy on "hackathons" and code burning sessions.

Late master's level/PhD: You won't even get a phone call until they've gone through your published work. They want to see journals, public solutions, examples of work, portfolios. They want to speak to your PI's. Your work IS your experience. If you can't take your work with you, that means starting from scratch, hundreds of thousands of dollars worth of research that you have to start publishing internally from day zero.
Before you even come on site for an interview, they want to have an idea of what you did on every project from the day you started your last job until the day you interview. Because they need to know if you can solve their problems based on problems you've ALREADY solved. Not having your specific projects accessible to future employers means they're basically rolling the dice on hiring you.
This isn't like undergraduate where there's a testing procedure to figure out your skills. Do you know how hard it would be to design a hiring test for a PhD level applicant in cognitive science? You would have to hire a post-doc cognitive science, esteemed professor specifically in the subject your applicant worked in, just to design the stupid test. Compound this with the hiring costs, and hiring a former Apple PhD could cost you a LOT compared to an employee with public research.
A PhD applicant/candidate is expensive not just because of initial salary but also time from hire -> time to productivity. It costs an average $20,000 to hire and train a $50,000 employee with a bachelor's degree.
PhD level hires, with a salary of around $110,000 to $140,000 could cost upwards of around $35,000 to $40,000 just to get them through the door on day one.
You think they're going to risk $35,000 to hire an employee that worked at Apple, the lead industry's last place company, who can't even show what he worked on or talk about his projects, just because of Apple's stock prices? Wake up.

Your blind loyalty to Apple as a fan is clouding your ability to see that working for a company and buying their products are NOT related at all. (I'm a huge huge Apple fan. MacBook Pro, iPad Mini 4, iPhone 6, saving up for a new Retina MBP)

 

[diipii]

Apple learned enough by sharing with Google back when Eric Schmidt was on the board. I'm glad they choose privacy as a focus instead of a free-for-all.

You've missed the significance of "free" and "all".

 

[ToroidalZeus]

But it's wealth accumulated primarily from Jobs-era legacy products and derivatives. It's almost like saying, a trust fund kid has business-sense because he's rich.

IF Apple was still pushing some fresh ideas based on the old "super secret" model you'd have a point, but innovation out of Apple Post-Jobs is muted and, frankly, on-par with the rest of the industry at best. It almost comes down to a GUI preference or the system one is locked into these days.

Or worse, the Apple's Beats deal was a bigger waste of money than Google's Motorola deal.

Based on their recent historical endeavors... I'm not buying that Apple is ahead of the AI game.

• Apple Maps - oops
• Ping - oops
• Facetime - limited
• Apple Music - oops
• Siri - oops

All had great potential but have been limited to Apple only or turned out to be far less than what they should have been. Based on this, I wouldn't trust anything AI Apple came up with.

One of the "dreads" in new technology is the thought that a company, military, or government is going to "black box" your idea and hide it from the light of day. No external collaberation. Limited validation. The exclusion of name or achievement recognition.

I wish Apple all the luck. I just don't think they can pull this off in any successful meaningful way. I hope I am wrong.

• ApplePay - limited uptake
• Watch OS - wait till version 2... or 3... or 4....

 

[kdarling]

I do think that Apple is falling behind in AI. Just look at Google Now, which is miles ahead of Siri. Apple is all about privacy though, which is what they are attacking. Apple's concern with user privacy is holding back the development of AI but it is keeping user's information safe. There's a trade off that Apple has made, some like it and some don't.

Yeah, there's something to be said for Apple's approach. However, it can require putting a lot of the intelligence on the device, instead of using mass server neural networks like Google does.

One reason I like the shared server approach is because the info is known to ALL my devices. When I do a search, it doesn't matter whether I'm on my Windows laptop, Android Phone, or iOS iPad... Google knows my history and tailors things for me.

It's also great for voice input, because it has had years to learn my voice from an aggregate of all my usages on any device. Apple's voice input also tries to learn, but it's compartmentalized by device for "privacy".

With Google, I can even pick up a brand new device, and it immediately has all the advantages of the years of learning. iOS has to start from scratch each time.

I fall in the middle. I love using Google Now because it gives me amazing results, but then I check Google Dashboard and it knows everything about me. I like Siri, and the fact that Apple isn't collecting a boatload of info from me, but it isn't very accurate at all.

"Knows everything" is hyperbole, of course

Heck, I wish Google DID know when I finally bought that item I'd been searching for, so it'd stop serving up ads about it.

And the Google Dashboard lets us view or modify the info, an ability which Apple does not give its customers.

 

[usarioclave]

I can tell you that people in the defense industry who work with machine learning and AI don't publish either. I'm not sure that's the model that Apple wants to follow though.

Has it actually hurt Apple? Well, they're behind in language translation, that's for sure. But if that technology comes to apple devices, even if packaged by third parties, that's not a loss on apple's part.

 

[3square]

Maybe an Apple employee can chime in (anonymously of course), but since I doubt they'll be honest, I'll go ahead and say it:
Apple operates at an almost isolated like environment for development and engineering, due to independent non-disclosure agreements. You can't have an employee working on the iPhone, talking to someone who works on the iMac at lunch about design aspects. Because how does the iMac employee know what he can or can't talk to his wife about?
So they all get separated.
I met an Apple software engineer. He said he worked in a small office, by himself, and only usually saw other engineers when he either needed guidance or a meeting about specs. But 95% of the time he's actually writing code, he's alone in an office.

This type of environment is due to Apple's very secretive operating policy for shareholders. Their stakeholders have a strong belief that keeping customers and public in the dark until announcements has a direct correlation on sales efforts.

While this might be true, the disconnection between their engineers and developers is a reason why iOS is starting to fall behind Android in a lot of major ways.

The other aspect of this that you don't realize is how terrible their separation policy is. Let me describe to you what happens if you're "fired" from Apple. Now, keep in mind, I'm not saying that someone who gets fired doesn't deserve to get fired. But you shouldn't just throw that blanket generalization around because not everyone gets let go for the same reasons, but NO one deserves to be let go like this. And if you think that this is okay, you're honestly a very inhumane person and I think the world is better off without anyone agreeing with you. Even Apple employees think this is harsh and cruel:
You'll be at work, and at a certain time, someone will knock on your door. There will be a security guard, your network will shut down and you'll have no intranet or outside web access. You'll be escorted away from your desk to be given separation papers and within minutes, you're outside of the building and that's the last you'll ever step foot on Apple campus.
You have no idea this was about to happen. Because if any of your superiors or coworkers "warn" you, they could have the same visit waiting for them (secrecy, NDA, etc). You might have an idea about it because of performance issues or other things, but your boss won't even warn you that you're on thin ice because of secrecy/NDA/etc. They don't want to risk you leaking information out of anger.
I've sworn never to work at Apple due to that, in an engineering capacity. I'll work for Google or even Microsoft, but Apple to me is very intimidating in that regard.

Sure, they are and will probably continue to be the #1 used device from a singular company but there's no denying that Android as an OS is starting to really make miles of progress ahead of iOS. It takes far too long to implement new features into their IDE. We still don't have a very good table/list design flow and Siri is pretty much garbage compared to Now and Cortana.

Siri is very very gimicky and most of all, she has no widely used predictive modeling. If you always ask her at 9:00 am every morning what the traffic is like between your home and office, there should be a very simple threshold to start giving you that information in the morning. And an easy opt out in case you don't want that. I could write the code for that in less than 20 lines. Why doesn't Siri have it?

1. don't join the marine and complain it's not like the boy scout. It's no secret, they will ride you hard.
2. maybe you're not using it right. my apple watch shows my commute every morning, and evening. my phone shows the apps i use the most at a certain time. i.e.: spotify during morning commute, hulu during evening commute. etc.

 

[currocj]

Well, I don't think I'd be giving advice to the most wealthy company in history, but that's just me.

There are several prior title holders of "most wealthy company in history". I'm not saying Apple is on a cliff about to fall, but the more you think you're invincible, the more you are open to attacks from new ways of thinking. It's the EXACT same reasoning of why Microsoft is no longer king of the hill, with different specifics and era.

 

[ToroidalZeus]

Me neither.

If Siri represents the pinnacle of AI, then count me out.

Computers are still incredibly dumb. Very little has changed in the past fifty years.

Nah.

For example they have software that can now digitize a book page or a resume and scan for keywords like an address.

 

[Hmac]

Apple's strict adherence to an environment of secrecy and privacy in regards to its software and hardware development has been suggested as a major blow to the company's potential for growth in the field of artificial intelligence. In a new article by Bloomberg, Apple was noted as a non-attendee at the Neural Information Processing Systems conference, an annual confluence of companies including Google and Microsoft where researches get together to discuss the progress and development of AI technologies.

In years past, Apple has attended the conference, but its emissaries were known to keep "a low profile" during the proceedings. In the midst of a mass sharing and celebration of discoveries and findings in the world of AI, many remain unsure of the Cupertino company's continued success in such departments if it remains attached to such strict secrecy rules. "They're completely out of the loop," said Richard Zemel, a professor in the computer science department at the University of Toronto.

The biggest threat posed to Apple due to this level of secrecy, according to Trevor Darrell, managing director of a machine-learning research center at the University of California at Berkeley, is the barrier to entry it creates for graduate students fresh out of college. The stagnant environment and closed-off atmosphere inhibits the company's employees from interacting with the rest of the scientific community, an issue that most potential hires may not be entirely comfortable with.
Earlier in the month, Apple acquired two artificial intelligence-related start-ups: VocalIQ and Perceptio. VocalIQ's natural language API hints at a more naturalistic version of Siri in the future, and even possible integration into the rumored Apple car project. Perceptio suggests the possibility of a more expansive and robust AI system for Apple, without the compromise of the company's in-depth privacy policies that pull Siri back from services like Google Now and Microsoft's Cortana.

All the same, Bloomberg's story suggests that despite Apple's enthusiasm to innovate in the artificial intelligence sector, the company could continue to lag behind in certain departments -- Apple Maps, for example -- due to its stances on secrecy and privacy.

Article Link: Apple's Culture of Secrecy Slowing its Artificial Intelligence Development

 

[JediZenMaster]

I think they need to not be so secretive about somethings like their plans for AI. I mean after all we all know that they have an army of lawyers that will go balls for the wall if there are any issues with stolen or copied IP.

 

[Hmac]

Sheesh. Another doom-and-gloom prediction from yet another high-profile media source criticizing the way Apple does business. It's been 30+ years now. Do these media idiots never learn?

 

[jwwetzel]

The folks on the outside miss the point completely!

Apple is not a company that exists to do basic research. Apple exists to look at the world, look at how to make it better, look at the state of the art, and create something beautiful with that. It is above all an artistic endeavor, less so a scientific or technological endeavor.

They don't need to attend a conference to stay at the head of the curve. And they don't need to publish papers and winky dink around doing basic research. They will happily let the scientists, grad students, and other folks handle that. They will winky dink around with the results of the basic research and try to create something with that.

When it is time to do something truly great, something truly positive and beautiful, Apple is the only company where this level of work can take place. It will always attract a certain type of person. They don't just want a grad student who is really good at stuff. They want a grad student who is really good at stuff, but more importantly, has an aesthetic appeal, wants to change the world, and HATES the status quo, and is ready to produce something that can be sold. After all, real artists ship.

 

[kerpow69]

Oh boohoo. Sounds like they're bitching about not being able to use Apple as their R&D arm. On the other hand, Siri is horrible. I ask it simple things all the time and it comes back with nonsense. For instance, I asked for directions to a hardware store near me, named Shur-Way Building Supply. Siri came back with something about not finding anything in India. I'm not in India.

 

[NightFox]

I think it's interesting that the majority of posters on here seem to place Apple's corporate interests above their own interests as a consumer - surely collaboration would advance AI more than isolated research behind closed doors? And don't think that MS, Google et al don't have stuff to bring to the table.

I guess that's extreme brand loyalty in action.

 

[ob81]

I think we are forgetting the fact that Apple has been taking top talent from companies, big or small, for years. They don't care about a barrier with college grads. Why? If they notice talent at another company, they are going to take it. They have actually shutdown companies because they took all of their key talent.

Meaningless article.

 

[trainwrecka]

Apple could very well be ahead of everyone else in AI but we all just don't know.

Exactly. "They didn't show up, and this is the only way progress can be made. Time to write a story about them being behind." As my mom would say, "Facts. Schmacts."

 

[Novus John]

They'll either stagnate or improve with time. Either way, if AI is still mostly research and academia driven, then Apple is swimming against the flow. No one in academia will hire a person with a four years gap without a proof of their excellence.