AI-improved Siri to launch at WWDC 2024, claims leaker

Just one week ago, the New York Times launched a lawsuit against Microsoft over its AI's training which "reads" New York Times copyrighted materials in order to "train" its AI. If Apple uses copyrighted materials in its training, then Apple would also become the target of lawsuits.

If the lawsuit is successful, which I doubt, (since web crawling has always been legal or unchallenged) then to get quality AI we'll all have to switch to using AI developed by communist China. Of course, Chinese AI will know nothing about any information that is anti-China.

Back in the early CRT iMac days the system help files were a veritable encyclopedia of relevant information about the machines. For example, it's where I learnt that the Firewire iMacs could act as Firewire repeaters even when switched off.

Fast forward a few years to the introduction of Siri and I was hopeful it would be pumped full to the gills of relevant information about the device it was answering to. Anything from hardware fixes, warranty support, spare parts, diagnosis, tips and tricks etc. The whole shebang. 

It never happened so I really hope someone is on the ball on this point and planning to make Siri artificially inteligent on all of that going forward. 

22july2013 said:

Just one week ago, the New York Times launched a lawsuit against Microsoft over its AI's training which "reads" New York Times copyrighted materials in order to "train" its AI. If Apple uses copyrighted materials in its training, then Apple would also become the target of lawsuits.

Traditional news organizations have finally clued into the fact that they no longer control the access points to the content they pay for the creation of and post online. That, combined with the internet ethos of "free information", has created quite a problem. The people who control the information access points (and profit from them) like social media companies and more recently, AI companies, are happy to exploit people's desire for free stuff and avoid paying news organizations anything for that information (which also happens to maximize their profits).

I'd hope Apple takes the high road on this and negotiates fair deals for the copyrighted material they use for training (similar to Apple News), but obviously if your competitors are able to undercut you by taking the low road, it makes that far more difficult. And yes, there's also the problem of countries which simply have no concept of copyrighted material in their legal system.

22july2013 said:

Just one week ago, the New York Times launched a lawsuit against Microsoft over its AI's training which "reads" New York Times copyrighted materials in order to "train" its AI. If Apple uses copyrighted materials in its training, then Apple would also become the target of lawsuits.

If the lawsuit is successful, which I doubt, (since web crawling has always been legal or unchallenged) then to get quality AI we'll all have to switch to using AI developed by communist China. Of course, Chinese AI will know nothing about any information that is anti-China.

That all depends on how Apple's AI uses whatever it's 'trained' on. The NYT's problem with Microsoft is about pretty straightforward plagiarism. Their AI scraped the NYT archives and regurgitates that information as though the AI came up with it on its own and pays nothing for the source material. That's the fundamental problem with most of the generative AI that's made such a big splash in the past year. It's not the emergent pseudo-sentient robot overlord that's had everyone wringing their hands. It's just massive data aggregation combined with a barely-competent linguistic algorithm that creates mash-up regurgitations of information owned and copyrighted by others. If Siri suddenly starts reporting the news without attribution and compensation for sources, that's a problem. If Siri attributes sources and those sources are compensated, as when you read information in the Apple News app, then it's a proper use of the technology.

If Apple's machine learning (their preferred moniker for AI) is really clever, Siri might be able to cite two or three sources on a particular news subject and tie it all together in a concise summary. It'd be like having a PA who gets up two hours before you do, reads all the newspapers and fills you in while you have breakfast. Even better, you could then ask Siri to drop specific articles used for the summary into a folder for you to read on the train while you head off to work. Siri could use sources already in its news app, as well as sources to which the end user has subscribed. For instance, the NYT is notably not in the Apple News app, but if you have a subscription, perhaps your Siri could have access to the content for your personal use. 

Apple has a long history of showing up to a tech party late, but having actually figured out how to use that given tech in some novel and indispensable way, even as Apple's competitors were "first" but made an obtuse mess of it. Giving Siri functionality like I described above would be a good example of how Apple might implement advanced machine learning with Siri, even as Microsoft's half-baked AI has been tossed out there to do mash-ups of plagiarized source material.

AppleZulu said:

22july2013 said:

Just one week ago, the New York Times launched a lawsuit against Microsoft over its AI's training which "reads" New York Times copyrighted materials in order to "train" its AI. If Apple uses copyrighted materials in its training, then Apple would also become the target of lawsuits.

If the lawsuit is successful, which I doubt, (since web crawling has always been legal or unchallenged) then to get quality AI we'll all have to switch to using AI developed by communist China. Of course, Chinese AI will know nothing about any information that is anti-China.

Apple has a long history of showing up to a tech party late, but having actually figured out how to use that given tech in some novel and indispensable way, even as Apple's competitors were "first" but made an obtuse mess of it. Giving Siri functionality like I described above would be a good example of how Apple might implement advanced machine learning with Siri, even as Microsoft's half-baked AI has been tossed out there to do mash-ups of plagiarized source material.

Apple wasn't late to the party.  They had Siri for +13 years, and still terrible.  Looks like they have no idea on how to improve it.  Now MS s ahead of Apple, have their datacenters ready for AI services and have been integrating AI in their apps and services with Copilot.  And what Apple has done?  If you ask me, Apple is the one with the half-backed AI, not MS.  We'll have to wait and see the announcements from Apple in WWDC and MS Build.