Apple finds difficulty recruiting AI experts thanks to tough user privacy stance
Apple has reportedly set about massively expanding its team of data scientists and artificial intelligence experts, but has found it tough to recruit top minds in the field as the company's privacy policies prevent researchers from getting access to valuable user data.
The size of the AI research group that backs services like Siri has "tripled or quadrupled" in recent years, according to Reuters. That does not include the nearly 100 positions that are currently open for experts in machine learning, a subfield of AI that deals with developing technologies to let computers make autonomous decisions based on their own analysis of previous actions or data.
In a bid to catch up with competitors --?particularly Google --?Apple has expanded its recruiting of academics, the publication noted, peeling back the curtain slightly to reveal how it uses some tools first created in university laboratories.
"In the past, Apple has not been at the vanguard of machine learning and cutting edge artificial intelligence work, but that is rapidly changing," Allen Institute for Artificial Intelligence CEO and University of Washington professor Oren Etzioni said. "They are after the best and the brightest, just like everybody else."
Apple is said to have met resistance over its stringent internal privacy policies, however. Researchers are able to access only a tiny fraction of the data generated by users of Apple devices, with some services --?like Maps --?flushing data after as little as 15 minutes.
As a result, Apple focuses its AI efforts on local analysis of data already on the user's device, eschewing the more typical approach of intermingling --?generally anonymized --?user data for research and analysis in the cloud.
Artificial intelligence is at the heart of Apple's next generation of services, and the fruits of the company's labor are already beginning to appear in iOS 9 and OS X El Capitan. Both operating systems bring with them enhanced spotlight search functionality, but iOS 9's improvements to Siri and new Proactive features clearly illustrate the path forward.
In iOS 9, Siri has at least some contextual awareness --?saying "remind me about this later" while writing an email will set a reminder to follow up on that message, for example.
Proactive, meanwhile, analyzes a number of variables to surface information the user may want at that particular moment, before they request it. iOS 9 will notify users when they should leave for their next appointment based on local traffic conditions, for instance, and can use data from emails to automatically create calendar entries or identify unknown phone numbers.
The size of the AI research group that backs services like Siri has "tripled or quadrupled" in recent years, according to Reuters. That does not include the nearly 100 positions that are currently open for experts in machine learning, a subfield of AI that deals with developing technologies to let computers make autonomous decisions based on their own analysis of previous actions or data.
In a bid to catch up with competitors --?particularly Google --?Apple has expanded its recruiting of academics, the publication noted, peeling back the curtain slightly to reveal how it uses some tools first created in university laboratories.
"In the past, Apple has not been at the vanguard of machine learning and cutting edge artificial intelligence work, but that is rapidly changing," Allen Institute for Artificial Intelligence CEO and University of Washington professor Oren Etzioni said. "They are after the best and the brightest, just like everybody else."
Apple is said to have met resistance over its stringent internal privacy policies, however. Researchers are able to access only a tiny fraction of the data generated by users of Apple devices, with some services --?like Maps --?flushing data after as little as 15 minutes.
Apple's AI strategy is front and center in iOS 9 with enhancements to Siri and the new Proactive features.
As a result, Apple focuses its AI efforts on local analysis of data already on the user's device, eschewing the more typical approach of intermingling --?generally anonymized --?user data for research and analysis in the cloud.
Artificial intelligence is at the heart of Apple's next generation of services, and the fruits of the company's labor are already beginning to appear in iOS 9 and OS X El Capitan. Both operating systems bring with them enhanced spotlight search functionality, but iOS 9's improvements to Siri and new Proactive features clearly illustrate the path forward.
In iOS 9, Siri has at least some contextual awareness --?saying "remind me about this later" while writing an email will set a reminder to follow up on that message, for example.
Proactive, meanwhile, analyzes a number of variables to surface information the user may want at that particular moment, before they request it. iOS 9 will notify users when they should leave for their next appointment based on local traffic conditions, for instance, and can use data from emails to automatically create calendar entries or identify unknown phone numbers.
Comments
The whole remiss sounds like an excuse to take the low road and digitally stalk users for profit.
It is much worse than that. iOS developers are prevented from accessing the full abilities of the devices due to privacy concerns. Apple used to have a really powerful Bluetooth SDK. Now it has been demolished because Bluetooth uses unique IDs. In order to hide those IDs Apple had to break the spec that they helped develop. There has never been an official WiFi SDK even though it is a public standard intended to be used by all developers. Part of the reason has to be because developers could use the unique IDs in the spec to track users and devices. Without full access to these basic digital communication features, developers cannot innovate. We would have had widely adopted mesh networks by now with a real WiFi SDK for example. What good is privacy if you have to wear shackles?
"Apple finds difficulty recruiting AI experts"
I thought that was referring to us
You beat me to the joke. I was going to quote "Apple finds difficulty recruiting AI experts thanks to tough user privacy stance" and then write, "They could just read our forum posts."
On the one hand, I admire and value Apple's dedication to privacy.
On the other hand, I recognize that it's hard to keep search/maps/Siri/etc. competitive with Google and others without substantial amounts of machine learning that require using anonymized queries to assess success.
So long as Apple keeps the queries to itself, I'm not sure that their use of them for this purpose would even violate their stated privacy policy.
From research to product there are some mundane mediocre obstacles such as people's security and privacy.
You have that reversed. If you ARE in shackles what good is privacy? Your freedoms and privacy are only assured if they are defended.
A few years ago AI was a joke and now all these big companies are spending up large on it.
????
Sounds like this is a good job candidate filtering tool, meaning it's easier to do the job when you have more data to build models and "innovate" in this field, but it takes real genius to provide similar if not equal or superior services when you have access to less information. Filter out the people who can't do their jobs without the easy (comprehensive) information and you'll have the people who are really going to shake things up with access only to privacy respecting data, because that's where the talent is required and that's where genius will be found.
You make an interesting point. Artificial intelligence really has less to do with "big data" collection and analysis and more to do with clever programming, neural networks and evolutionary models. The invasion of privacy angle is a false argument, one more likely to be made by a spy agency...or Google.
Bingo. It's a false dichotomy. In the pre-Google world, AI meant machine intelligence, machine learning, and natural language parsing. In the Google world, AI now means creepy data collection about human test subjects? Are we building HAL9000 or f-ing SkyNet, An Alphabet subsidiary?
At 60 posts you're not at expert level yet.
Bingo. It's a false dichotomy. In the pre-Google world, AI meant machine intelligence, machine learning, and natural language parsing. In the Google world, AI now means creepy data collection about human test subjects? Are we building HAL9000 or f-ing SkyNet, An Alphabet subsidiary?
It's because no one yet knows how to do real, thinking, AI, so they are trying to put lipstick on a search engine. And search engines need big DBs.
I would rather have Apple kill Siri than lose privacy.
If AI experts need private data, then they aren't that bright to begin with.
This is silly. All companies have to do is ask when they want your data and tell you exactly what they're doing with it. I personally have zero concerns that Apple will improperly use my data so I have no problem sharing it with them if it ultimately makes for a better user experience.
This is silly. All companies have to do is ask when they want your data and tell you exactly what they're doing with it. I personally have zero concerns that Apple will improperly use my data so I have no problem sharing it with them if it ultimately makes for a better user experience.
I agree. Here it is from the man himself.
And another great article on the balance between privacy and user experience;
http://dcurt.is/privacy-vs-user-experience
"The real issue that Apple is trying to address is not really privacy, but rather security. Though Google has all of my data, it is still private. Google does not sell access to my data; it sells access to my attention. Advertisers do not get my information from Google."
"The vast improvements in user experience far, far outweigh the potential security risks to private information."