It's a shame that Apple doesn't have an industrial research lab; it certainly has the funding to create a lab that would rival Microsoft Research or perhaps even legendary labs like Xerox PARC and Bell Labs. Apple used to have a lab: during the "interregnum" years of 1985 to 1997, Apple had a fantastic research group called the Advanced Technology Group, led by the late Larry Tesler. In some ways this group was a sort of spiritual successor to Xerox PARC; Larry Tesler was ex-Xerox, and the legendary Alan Kay (of Smalltalk fame) and Don Norman (who wasn't from Xerox PARC but who is a legend in usability) were involved in this group. This group worked on many interesting and important technologies, such as Quicktime, AppleScript, OpenDoc, HyperCard, speech recognition, and more. Even though the Dylan project did not come from the Advanced Technology Group, it is another example of interesting work that came out of Apple during this time period. Even though commercially Apple struggled during the latter half of the interregnum, Apple created some amazing technologies during this time period.

Of course, the Sculley/Spindler/Amelio Apple seemed to be far more open regarding research than the Jobs- and Cook-era Apple. Jobs closed down the labs in 1997 and help institute Apple's famous culture of secrecy that persists today.

Has the culture of secrecy benefited them at all since the launch of the iPhone?

I think the answer to that question is yes.

I'll argue that AI leadership is not limited to just research, and an under appreciated aspect of leadership is making actual things useful for people (e.g. applying AI/ML technology to problems, and making those solutions common or ubiquitous).

It would be hard to be leader without having leaders in the field at your disposal. There are literally 20,000 papers getting published each year. While stunning gains are being reported every month, it would be very hard for you to identify good implementable papers with state of the art results unless you are active in the field. Many of the techniques are amalgamation of other dozen techniques. So its even more harder to add/subtract from a given paper to productize the results. Keeping up with these papers and techniques is literally a full time job which is why you need researchers at your disposal if you are working on these problems. Unfortunately, researchers are almost never willing to just give up research and become your next engineering dude. They are in the field because they love research, freedom and can publish. So the viable way is to have a research lab which you can tap into on demand and have options for rotations. Many people believed that this can be done without having your own research lab but that doesn't often work out. If you approach a researcher in academia for consulting, you have to count on them being free from academic duty, grad students and other commitments. They will typically do 80/20. If you have your own research lab, you can turn this around to 20/80.

There's more to leadership than research and pushing state of the art in papers.

There's also a lot more to making useful AI/ML-powered technology than having good models.

Defining and limiting the field to just research and models is a common pitfall I've seen.

Apple has fallen way behind in software/apps, and maybe they’re ok with it. The hardware and OS are good, and some staple apps for creators are good. But at everything else they seem really lethargic or absent. For example, as a web developer Apple has had zero impact on my tooling and knowledge. But their laptop is nice and I have one, which may be good enough for them.

Apple has had some impact for me on web. I can ship an ogg/webm universally in <video> and <audio> tags, but Safari on iOS is the only one that demands a mp4.

FWIW, I used a MacBook Pro for work (2018) and the wrist rejection was bad enough to make it unusable. Whenever I typed, it would constantly move the cursor around, usually with a click that would change where my text was going, no matter what settings I adjusted.

It was less of an issue with the older MacBook airs, but the Pros now have a giant oversized trackpad that you have to touch when typing unless you hover.

What is wrist rejection ?

Recognizing to disregard input from your wrist/palm when it's resting on the iPad surface whilst one is using Apple Pencil.

...Psychosomatic cognitive dissonance?

I don't own or use Apple products so this takes more explaining for me.

You're typing on your laptop and you don't want your wrist/palm to accidentally trigger anything on your trackpad, so you employ wrist rejection.

That’s an interesting take. Does Apple really lag? I think quote-unquote AI in SV has pretty much stalled in the past 3-4 years. There’s been a lot of volume in ML but nothing groundbreaking.

That's not true at all. AI is everywhere. When Gmail completes your sentence, auto-playlist gets created, camera detects a pet, you take night shot of a city, grammar suggestions are shown, wrong keystrokes auto-corrects, voice mail is transcribed, phone call is responded by quick text, your sleep is analysed, heart rate is monitored on runs, news articles gets recommended, spam email gets filtered, device idle detected... I can go on.

Most of these problems cannot be simply solved by just throwing in best engineers. No amount of classical algorithms you learned as CS major is going to help you implement the best solutions for these problems. The state of art solutions to these problems requires intense narrowly focused researchers who have studied these problems for many years, knows which 10% of the papers even worth looking at and pro/cons of different techniques. Something as benign as running neural net on phone hardware is intensely researched subject and your implementation can be literally 10X to 100X better at speed and power consumption if you have kept up with the field.

None of those things was solved in the last 3-4 years though.

Some might now be done via AI, but is it done any noteceably better than 3-4 years ago? I certainly don't trust my camera to adjust itself, or the sleep analysis to be any good, or the article recommendation to not recommend based on outrage, or the spam filter to work at all.

That's the whole point. If you are still using classical solutions for these things that are years old, you are far behind best experience and the competition. As an example, noise removal in low light images has advanced tremendously each year with stunning gains. Same goes for grammar correction and auto-correct. You can see the difference in Siri and Google Assistant that is literally an order of magnitude. Siri even has trouble doing proper voice recognition and as Apple does not even know how to do search, its question answering skills shines only for highly curated tasks. First thing I've always done on my iPhones is to turn it off. However, Google had been amazingly improving these stuff every year. The end result is that while I do use iPhone, I tend to use mostly Google services and when something is only available on Android I get a tremendous itch to switch.

It sounds like you are just advocating the corporate tech transfer model of advanced research in a nascent field, but maybe Apple is philosophically opposed to doing that. Indeed, it's not that they don't have funds for scientific research. It's easy to say Siri is inferior or whatever but that contains such hidden assumptions about what really is valuable for consumer tech.

> You can see the difference in Siri and Google Assistant that is literally an order of magnitude.

How is this measured?

To compare that list to apple, my Mac often (but not predictably) complains that things like "the" isn't a word. Literally worse than a lookup against /use/dict.

My 2c is that goodfellow is probably working on their self-driving project.

i dont have any inside insight but id be surprised if thats the case. goodfellow's line of research has been adversarial learning [1].

[1] https://scholar.google.ca/citations?hl=en&user=iYN86KEAAAAJ&...