Nothing says “We’re confident in the software we’re selling” like willing to work for exposure in hopes that somebody shills $20 for a subscription.
Exposure = They get to keep the data they get.
Data = money
They’ve found a way to make it work I’m sure.
Apparently they won’t be collecting data.
Privacy protections are built in for users who access ChatGPT — their IP addresses are obscured, and OpenAI won’t store requests. ChatGPT’s data-use policies apply for users who choose to connect their account.
https://www.apple.com/newsroom/2024/06/introducing-apple-intelligence-for-iphone-ipad-and-mac/
They pinky promise not to collect the data
I mean Apples the one saying it. I doubt OpenAI wants to piss them off.
How would Apple technically know that nothing is stored? Even if IPs are hidden, the questions and answers are still useful for OpenAI.
They probably have a deal similar to DuckDuckGo:
As noted above, we call model providers on your behalf so your personal information (for example, IP address) is not exposed to them. In addition, we have agreements in place with all model providers that further limit how they can use data from these anonymous requests that includes not using Prompts and Outputs to develop or improve their models as well as deleting all information received once it is no longer necessary to provide Outputs (at most within 30 days with limited exceptions for safety and legal compliance).
Yeah there’s definitely a contract, but open ai could determine it’s more profitable to void the contract and pay for lawyers and a settlement. Probably unlikely though to be fair.
I don’t believe that at all! There’s some kind of catch.
Even if they don’t sell the data I’m sure it’s valuable for their models.
This seems so obvious I’m amazed Apple isn’t charging them for the “exposure”
Apple needs them just as badly as OpenAI needs every iPhone user. They’re horribly behind in this “AI” bubble.
With the sheer amount of money that the rich are throwing at OpenAI via investment firms, they don’t need nor want to charge imo. The fact that they’re being built into Apple’s ecosystem and are getting name-dropped to people inside of iOS is kinda what their investors want.
It’s the age old “walmart opens and operates at a loss for 2 years to force others out of business, then jacks the price” model.
Investors want them to cement this as The AI company & brand so that once it gets giant and starts to be profitable just by being the biggest gorilla in the room, the shares they bought are worth more.
So what I’m trying to say is that our version of capitalism is perfect and makes lots of sense and is in no way insane and degenerate.
That might be their internal reasoning but Apple will very quickly move to have these capabilities in house. Apple has been working on machine learning for a while but they don’t collect data so they are unable to build these LLMs.
For now it makes sense for Apple to leave the liability of basing these LLMs on copyrighted data. If OpenAI losses those court battles, they take the hit for services rendered to Apple. None of that liability transfers to Apple.
Meanwhile, Apple is going about this the Apple way by encouraging developers to integrate their apps into new frameworks being added. This gives them access to user data directly from the source allowing them to build personalized models.
These models will likely be far more useful to the day to day mundanity of life than the hallucinogenic encyclopedia that is ChatGPT.
Imagine all the training they’ll get from random queries off of the millions of Apple users who don’t even know what chat GPT even is yet.
Supposedly they’re not allowed to use any data obtained from this for training purposes, at least according to the mkbhd video
It’s like telling your water to not run down the drain
I guess it deepened on what the onboarding flow looks like for newbies. The per-query alert is pretty minimal.
That said, my original point is that training data gathered from queries is probably not valuable enough to offset the costs of unpaid GP4 query compute for the biggest smart phone manufacturer on earth.
The data is valuable, but for GP4 access, OpenAI would rather scrape chat forums, sell integration licenses, or sell pro licenses to offset all those damn Nvidia chips.
Maybe OpenAI is hoping the cost of a couple million queries within a month will be offset once they start charging for it within a month or two.
I doubt that’s the play. OpenAI is probably locked into a service contract, and Apple had built the platform so different 3rd party LLMs can be swapped out by Apple or the end user. So if OpenAI breached the contract, Apple could go with a different default model.
You are the product.
I don’t know with this one.
OpenAI isn’t allowed to blindly scrape iPhone or iCloud user’s data. They can only learn from whatever data the user submits in a query. And Apple forces users to consent to every single query that is sent to OpenAI. And that query compute is expensive.
Given that Apple built in a way for GPT pro licenses to be used, my guess is that selling the subscriptions is the real business angle.
Open AI will get data from those queries, but since it’s a query, it requires significant compute. Paying for compute for the world’s largest smart phone manufacturer is going to hurt if they can’t monetize somehow.
- Deploy a Turing-test beating interface in all Apple devices
- Appoint NSA to your board
- …
- PROFIT
100% correct.
At the moment there are tons of Ai companies all trying to be The One that everybody will use.
But Google has Gemini and it has Android. Android has 70% market share worldwide and is offering Gemini for free to every user. That’s MASSIVE exposure.
For openai to get on Android the user has to install it as an app. It’s VERY difficult to get your average user to know what openai is and why they should use it, never mind getting them to install the app.
So to be the default Ai on iPhone is a HUGE deal for openAi and gives them massive exposure over the competition.
Google pays Apple billions to Apple to be the default search engine so openAi not having to pay anything is actually very surprising to me.
I think the only reason Apple isn’t making them pay is because Apple plans to offer other Ai services in future.
That’s a really good point.
OpenAI is also pushing pretty hard to get partnerships with Samsung. If they had Apple and Samsung, they’d have half of the phones in the world.
I wonder how this will end up working. I want to use chat gpt without an account and while logged into a VPN and to have unlimited requests… would be nice if this was the solution to this.
Check out ddg’s ai chat.
I need to use this more often. Good call.
Apple just had the negotiator of the year right there, or OpenAI is really that bad.
It’s especially bad knowing that I’m this setup, all of OpenAI is replaceable in Apple’s code by changing a single line of code and pointing to a different API.
My guess is that Apple is just able to show OpenAI how good they are at upselling other 3rd party apps into paid subscriptions.
I bet the bargaining chip is, “you can sell GPT Plus, but we won’t take a 15 to 30% App Store cut if you let us send requests to GPT 4 when the dumber Apple LLMs don’t cut it.”
It will be interesting to see the upgrade numbers in a year. Do most people not care about AI, or will the user base be wary? Current numbers show around 77% uptake on iOS 17.
https://wccftech.com/ios-17-adoption-rate-lowe-than-last-year/
My hot take is that the new iMessage features will push adoption rates really high. People in my household want the beta for that reason alone, and I’m having to bat them away because this is a buggy DB1.
A version is Siri that isn’t shit is also a big reason. But that is not coming until 18.1 or 2. So my money is on text effects and stupid emoji reactions being the initial upgrade driver.
