blastdoor

Apple researchers take aim at AI hallucinations and true conversations

blastdoor

July 25

Apple employs very smart people and is definitely a leading technology company. They are not behind in that regard.

But they are behind in integrating those technologies into a compelling AI product — that is, a truly useful product that people and organizations are willing to pay for.

They can certainly catch up, but it’s delusional to assert they aren’t behind.

Doom and gloom reporting on Apple Intelligence continues to ignore Apple's playbook

blastdoor

July 23

Wesley_Hilliard said:

blastdoor said:

Wesley_Hilliard said:

I think there's some truth in the middle here. AI is a tool, and can be useful, in certain circumstances. I've never dismissed it as a passing fad, but I do think the hype around it is overblown nonsense from those seeking investment capital. I use AI every day (Apple Intelligence) and I benefit from it. Apple is leading the market in creating powerful, on-device, private, and secure models while also allowing users private access to leading AI platforms. It'll prove to be an incredible combination over time.

I do not subscribe to the idea that AI will take over or become sentient. It's going to make humans more efficient at certain things, and render some jobs redundant. But not because the AI is doing the job, but because it'll take less humans to do the same work. The writer worried about losing his job to AI shouldn't be, because even if you write with AI, you'll need human intervention to give it soul and reason -- which AI will never have.

That's why it's so funny to me that people see Apple as so behind. It's laying the groundwork for the future of a cooperative AI ecosystem built on Apple platforms with Apple's rules and values, and because it isn't complete this second, it somehow means they're lost in the woods. As with nearly every Apple endeavor in the past 30 years, I wouldn't bet against them.

If the only AI you use is Apple Intelligence then you're in no position to assess whether Apple is behind. I use an enterprise license for ChatGPT almost every day. Apple offers nothing like it -- they absolutely are behind in terms of offering a product that competes with what ChatGPT can do today. In terms of raw technology I agree that Apple has a lot going for them. But they have yet to create an AI product that is as useful as ChatGPT.

Accurately recognizing that they are behind is not the same thing as betting against them. I also would not bet against them. But I can open my eyes and see that for the moment, they are absolutely behind in this market with respect to actual products that are useful to people and worth paying for.

Someone can assess a vehicle is fast without driving it. What are you using the enterprise ChatGPT license for? What product are you using that you believe Apple should be offering?

But you can't assess the difference between an EV and a combustion engine without ever driving the eV. It's a qualitatively different experience, and using ChatGPT is a qualitatively different thing than what Apple Intelligence is today. Actually, it's more like the difference between a go cart (Apple intelligence) and a Tesla Model S. If all you've done is drive go carts, you have zero clue what a Model S is like.

I can ask the o3 model to write an R Shiny app (what I'm doing right now) with so-and-so features and it does it. I can then iterate productively to refine the app. I can ask it why it did things and explain to me how various aspects of the code work, so that I learn more (I've used R forever but I'm new to Shiny). This effectively replaces a research assistant or programmer for me. It's a huge productivity boost.

For another example -- earlier today I asked ChatGPT whether there's a connection between conducting a fixed effects meta-analysis using weights to account for error covariance and conducting a principal components analysis. It explained the connection and then, based on remembering an earlier conversation, suggested how this connection applied to some other work I was doing. If I asked Siri anything like that the answer would be "here's what I found on the web"

Apple Intelligence translation for users, new AI tools for developers coming at WWDC

blastdoor

June 7

https://daringfireball.net/linked/2025/06/07/swift-6-llms

Gruber talks about the productivity disadvantage for developers using Swift 6 because Apple doesn’t have AI assistance for developers and the lack of open source swift 6 code means other AI developer tools also don’t support swift 6 well.

apple is falling behind in a serious way here

Tantalizing details of Jony Ive's AI device leak after OpenAI meeting

blastdoor

May 22

bloggerblog said:

Remember OQO inc, ModBook, or General Magic?
There's a pattern where Apple ex-employees failing on their own, not saying the pattern won't be broken. But that it exists

I think there's a very distinct possibility that it will fail hard. Even if it ends up being a good product, Apple will likely copy it within a year or two and take over the market.

If this does end up being a tricorder type device, privacy/security concerns will be immense. Apple's strong brand in privacy/security combined with PCC can give them a leg up. Also, I wonder if it will turn out that a separate device really isn't necessary for people who already have multiple apple devices. Maybe OpenAI sees a need for a device because they currently sell no devices.

US and China temporarily lower tariffs to start trade negotiations

blastdoor

May 12

The difference between a 145% tariff and a 30% tariff is that with 145% shelves would be empty, with 30% the shelves won't be empty but the stuff on them will be a fair bit more expensive than before. Remember, people lost their minds when prices went up 7% over the course of a year under Biden.