Certainly the copilot button in ms paint did nothing to attract the clawbot ecosystem to windows
OpenAI's ChatGPT is AI consumer software and is a hit, albeit mostly free tier users.
And yes, ChatGPT is a hit but who will subsidize the hardware for freeloaders, Google's (cheap to run) AI is good enough now that I don't need to move over to ChatGPT for simple answers, thus the Google moat will probably remain intact denying OpenAI the search revenue stream all whilst OpenAI proposals/trials to add ADs were met with annoyance.
AI where useful is becoming a commodity, Apple did the correct thing in waiting and using the commodity parts and we're otherwise also quickly heading to the bubble's pop, HN even censoring articles on the topic sure seems to be an indicator that those in power are afraid.
They have no ground to make up on AI, and changing their operating system to center on AI would piss off every iPhone user I know outside of tech, and probably half of them within tech.
[1] https://www.macrumors.com/2026/05/13/apple-q1-market-share-g...
Apple doesn't leak much but there has been coverage of this:
https://spyglass.org/apple-ai-fail/ (April 2025)
https://www.theinformation.com/articles/apple-fumbled-siris-... (paywalled)
https://appleinsider.com/articles/26/06/07/one-fateful-meeti... (2 days ago)
Well, the results[1] are[2] actually[3] in. Samsung of course did do that and the results are what you'd expect.
So in a sense Apple 'could' have released what they had, after all Samsung and others have, but almost certainly not at the level of quality Apple expects. In which case arguably not releasing until it is capable of reaching that quality bar is the right call. The wrong call was announcing it in the first place when it wasn't ready.
[1] https://www.reddit.com/r/samsung/comments/1b4zc1j/new_ai_tex... [2] https://tech.yahoo.com/ai/articles/awful-galaxy-s24-feature-... [3] https://www.androidauthority.com/im-tired-pretending-galaxy-...
Of all the things they could build, why must they pick this future...
Wrong. It definitely wasn't 50%. This number seems to grow each time I see it referenced. The Stasi directly or indirectly employed about 2% of the population, which is still huge. But /The Lives of Others/ takes a lot of artistic license. The true levels will never be known, but the largest and most widely quoted figure is 1 in 6.5 or about 15%. That derives from one historian's estimate, which was that at the upper bound, 1 in 6.5 people had in some way made a report to the government that in some way made it to the Stasi. I'm sure 15% of the people in any developed country have called the police at some point in their life.
This is also assuming no duplicates, you really think the Stasi could uniquely identify and disambiguate informants at this scale? And that every Stasi low level officer tasked with recruiting new informants or else actually recruited new informants instead of making them up and keeping the payouts for themselves?
And because I have to say it: authoritarian surveillance is bad, the Stasi was bad, this is not an apology or minimization, but a correction of historical facts.
While I don't necessarily disagree with their vision but if implemented like "Copilot for Windows" I don't see me or anyone wanting to go anywhere near it.
Apple being slow is just fine, at least they didn't launch "Copilot for Mac".
Sometimes the lack of certain feature is the feature.
Is this not what Apple Intelligence is?
Sometimes it's a curse. Apple might be a 2-3 trillion dollar business right now, if they didn't refuse to sign CUDA drivers for their ARM servers.
I think agents are scary and complicated and dangerous enough that it is genuinely scary to give an agent an instruction like go buy this ticket. It’s okay and apple can easily simplify and eventually win. The mainstream hasn’t really started using agents yet and no one has come close to delivering a platform that will get them there.
Once again, early 1990's General Magic looks prescient.
They were working on smartphones with agents capable of completing remote transactions before we had wireless data networks.
> General Magic: The Greatest Tech Company You’ve Never Heard Of
https://m.youtube.com/watch?v=tuFl4WEXBrk
https://www.wikipedia.org/wiki/Telescript_(programming_langu...
These ones also seem really weird because the baseline is most often someone using the iOS app to do the same thing, and the agent demos are usually slower in addition to being riskier. One of the Chrome demos had someone buying groceries at pretty hefty markup, which seemed to be targeting a narrow demographic of people who a) don’t worry about paying 50% more for produce and b) can spend time writing a prompt but not 30 second opening an app and just doing it with zero chance of getting scammed.
you will be surrounded by an ecosystem of
devices, none of which stand alone, but are
more like portals to interact with your agents
And then a minute later I could ask it "Is copilot done yet?" and it replies "No, looks like it is still working on it". And then a minute later I ask again. It replies "Yes, it finished. It changed chart.py and styles.css. Do you want me to tell you what specific changes it made to the files?".
But it looks like none of the chat agents with voice interface have such a connector at the moment? An SSH connector would be the most useful. But a "GitHub Codespace connector" or something like that would also do.
I wonder if that will be a missing piece for long. If so, I would build an agent with voice mode and ssh connector myself. But I guess it should come out from the big guys any moment now?
A verbal diff sounds practically useless. Does it first read out the entire left-hand base, and then read out the entire right-hand target? Does it say loudly "REMOVING ... ADDING ... "? How would it read out something like Struct->Field? This seems lower fidelity than a visual confirmation, and I just don't think that voice commands make sense with this kind of work.
"It changed the plot function so it takes another parameter called linewidth. It also added an input field in the stylecontrols section where the user can ...".
Every time I try to take a hands-off approach to the code like this, I come to regret it later. The code ends up bloated and labyrinthine. When I let it grow unabated, it becomes gradually more difficult for the LLM to understand the intended structure as the project becomes too big for the model to keep the whole thing in its context.
How would you detect the
presence of bugs in this
scenario?
How would you make sure the LLM
isn't adding yet another
useless, redundant function to
the code base?
Please don't tell me that never happens-- I've had one just in the last week and I use both OpenAI and Anthropic foundation models.
In fact, I usually let multiple LLMs implement the same feature, and then I compare them. I even run my own arena in which I calculate Elo scores for LLMs from my perspective of which one implemented features better.
Having the ability to control code agents via voice would not take away my ability to do that. But I think in the future, that will become less and less necessary. If we look back at this conversation in five years, it will look very archaic, and we will be used to having superhuman AI do everything for us. In 10 years, it will sound like a strange idea that humans were once fiddling with code to improve the quality.
Something something taking the crafts and the man out of craftsmanship to just get it out the door as quickly as possible.
All jest aside I mostly agree with you but I'd tack on another 20 years for a total of 30.
Though in this technological jump I don't think people are as excited (understandably) as when the teletype came on scene. I too like the potential but dislike the whole discourse around it, the ethics involved and the way it's deployed. Such is life I suppose.
Just not doing it for me. Think I'm gonna stop reading anything he says.
Edit: missing words, thinking faster than typing
I don't care about a twice a week podcast about the NBA and national parks, or the other 5? podcasts about random stuff.
My problem is part style, and part content. Stratechery reads like it's written to be narrated - rather than exist first as writing. There's verbosity, pauses, long sentences, etc. And then you listen to the narration it makes sense.
But that complexity makes reading harder. Not saying everything needs to be 5th-grade-level, but complexity isn't required. Paste a Stratechery article into Hemingway Editor to visualize my point.
The stats below:
Readibility - Post-Graduate (aim for 9)
26 of 44 sentences very hard to read
8 of 88 sentences hard to read
31 weakeners
6 words with simpler alternatives
What a chore to cover, and that's without commenting on the ideas/concepts in the content.
I'm sure some folks like this writing style but I don't. And try hard to write my newsletter and other prose with far less complexity.
He was disappointed in the Apple vision pro for just being an entertainment device (it seems like you two agree there?)
And then the interviews by media of tech should be viewed as an iterated game. He can ask interesting questions for an analyst, but he (and Nilay) do depend on access and that fundamentally constrains what types of questions they can ask if they want continued access
> Just not doing it for me. Think I'm gonna stop reading anything he says.
Pretty sane take tbh
But when it comes to anything around consumer behavior, individuals, etc, i.e. the average family in America, he is often completely and utterly wrong in all his takes and predictions. In fact, so wrong it's often laughable, and amazes me that he is so confident in his predictions.
Also, in the podcast I've noticed that he talks almost every podcast about his "hits", i.e. his times in the past where he predicted something accurately. But never, ever mentions the times where he was completely wrong. He's like the dictionary definition of confirmation bias (or survivorship bias).
It's like he's gotten overly confident (or a little arrogant) as he's become more of a tech celebrity, to the point where he thinks he's some sort of Nostradamus now and doesn't recognize his weaknesses or failures. And I've personally stopped listening to the podcasts as much as it's getting a little tiresome.
BTW, I also noticed how often he is wrong on deep tech topics, e.g. his explanation of IP addresses and routing in one podcast. It's like he thinks his business knowledge + Claude is enough for him to authoritatively discuss how technical systems work, and he often is mistaken...
Yeah might as well cancel your subscription if you’re not gonna read it
For stuff that can't be run on phones, some of it will be run on Apple's servers, which I'm assuming Apple is eating the cost of for the time being.
Stuff that needs heavy reasoning or external knowledge will be processed by google, in exchange for $1 billion a year. However Google already pays Apple $20 billion a year for google to be the default iOS search engine, so you could view this as just changing to google paying $19 billion a year instead.
> PCC delivers a powerful server model without compromising privacy: data is never stored, used only for the request, and independently verified. It's integrated with the OS and iCloud, so there's no authentication or API keys, no token cost to developers, a daily per-user limit (higher with iCloud+), and eligibility for apps under 2M downloads.
Source: summary on https://developer.apple.com/videos/play/wwdc2026/319/
I haven’t seen any information about what’s happening with apps over 2M downloads, who graduate from the Small Business Program. https://developer.apple.com/app-store/small-business-program...
Some of it is free on-device. Some of it is free & rate limited per day. They mentioned in the WWDC infomercial that users with iCloud+ (the storage tier subscriptions, Apple likes to throw random things in with that) will be able to get more uses per day. And some of it developers will pay for.
And even if the assumption turns out to be wrong, they can just scale down and serve dumber and cheaper models. Shrinkflation is not a novel idea.
You might re-title the article instead, "The iPhone holds its ground", and it would be a more realistic title. But perhaps garnering less clicks.
I've always thought Ben Thompson is strong on enterprise and b2b topics but super weak on everything consumer related, he simply doesn't seem to understand consumer behavior (he has zero empathy or ability to project his mind into the average person's mind)
E.g. Ben was sure iPhone air would be a massive hit because he himself loved it. (It's struggled as people don't like the smaller battery life).
Ben was sure the Vision Pro would be a huge hit because he himself loved it. (It was a total failure as the average person doesnt want to pay huge amounts for a ridiculous looking dork helmet).
Ben raving about Meta's hand controller which he was sure was going to be the future of consumer electronics (The Neural Band). He was discussing how you could use it while your hand is in your jeans/pants pocket. Not quite thinking about how this would look while you're sat on the subway with someone sat opposite you.
Ben discussing how the future of watching sports is in VR. Not considering how weird it would be to go to a friends house to watch the game and everyone has their own VR headset. Also not considering the fun of watching sports is doing it with other people.
Basically, he has a huge issue with extracting his own liking of techy products to the average consumer who are basically nothing like Ben Thompson.
I know people are desperate for a Siri that works. The convince of just talking to your phone is priceless. If Apple gets this right, this is a huge deal – which it seems they are on the right track.
People are still talking to Siri for basic stuff like timers and alarms because it works, doesn't need an app, works when phone is locked or even away from you. If this works for more complex tasks like texting and general questions Apple will have the upper hand over Meta and Google in this new way of using computers/internet.
Apple also took a very clever approach for Capex and general AI strategy. Everyone knows that the best intelligence will eventually become a commodity and Apple decided to step aside from this expensive experiment. That's worth pointing out too.
The Air was interesting because everyone I've seen hold it, loves it. But, everyone also loves battery life and the best camera more. The Air is proof of that (similar with the mini lovers).
Just because you like something, doesn't mean it will succeed. These people will more likely using some sort of industry knowledge to form conclusion which conforms with their bias.
On the flip side, just because you hated something doesn't mean it will fail. There are plenty of Apple haters who will write things that seems to make sense but completely misses the mark every single time.