transpo1
macrumors 68000
They never should have tried to build their own LLM but the R&D expenditure could have been worse. Guess it's all working out 😀So no LLM SIRI from Apple itself
They never should have tried to build their own LLM but the R&D expenditure could have been worse. Guess it's all working out 😀So no LLM SIRI from Apple itself
Came here to say thisWell, let's hope no data goes to Google. If so, it's a no from me.
You're assuming people will still want and be using AI in 20 years the way email stuck aroundExactly the same as when Google search became the default search 20 years ago knocking out all the other search engines. This will be for another 20 years.
So no LLM SIRI from Apple itself
They wasted too much time doing dumb stuff like re-inventing the wheel with Swift and all of their system frameworks instead of working on more important things. They'll probably never be able to catch up.The company isn't capable of it
I believe it's a copy of Google's model running on Apple's own server hardware. Google's model is open source and free, I run it on a Windows box with a beefy graphics card using LMStudio.I don't like this at all. Google is a tracking hound, and Apple should know better.
But it’s not going to be Gemini, it’s Siri/Apple Intelligence that’s using Google’s models instead of Apple’s. Apple is reportedly paying a billion dollars a year for this. It is going to run on Private Cloud Compute, not Google’s servers.I don't understand people sometimes. There is no using Gemini without Google getting some data in return. Unless Apple is cutting a pretty massive check for a version that runs exclusively on Apple private cloud compute.
What drives ML development is data so people expecting Google to not have access to the data you provide by using their product are living in lala land.
People were writing Google’s AI obituary last year and saying it could never catch OpenAI and here we are. Fortunes change in an instant.They wasted too much time doing dumb stuff like re-inventing the wheel with Swift and all of their system frameworks instead of working on more important things. They'll probably never be able to catch up.
While that makes sense on the surface, the EU will likely run into a massive technical wall here. There is a fundamental difference between a 'surface-level' integration (like the current ChatGPT opt-in) and the core 'Apple Intelligence' architecture.The EU will force Apple to offer several LLMs to choose from just as they did with browsers, search engines, or app stores.
No, people are complaining because Google is the enemy of all human individuality.Because people have a huge emotional investment in Apple exceptionalism. Any suggestion that Apple has had to be pragmatic when solving a complex problem by partnering with other companies rattles that precious little construct.
Notice how the boilerplate response lately is always "well, Apple should do [x] themselves, then." Which is unmoored from the reality of how increasingly deep and wide challenges are solved. Apple gave Nü Siri the ol' college try. It went nowhere. If anything, they should've realized it was a lost cause far earlier.
More broadly, Apple in general lately waits far too long to abandon stupid or failed ideas. The advantage of their deep, deep pockets is not pouring endless money on go-nowhere efforts, but rather the freedom to walk away and use someone else's better ideas instead.
Today’s iPhones cannot be turned off, that choice was taken away from us.As long as I can keep it 100% disabled, idgaf
They can just replay the 2024 iPhone 17 Apple Intelligence vapourware stuff.Expecting Apple to talk more on Siri/AI/Apple Intelligence at WWDC.
Combined with...1. it’s Siri/Apple Intelligence that’s using Google’s models instead of Apple’s.
2. Apple is reportedly paying a billion dollars a year for this.
Because the math doesn't make sense if you take #1 and #2 together. There is no way Google agrees to give Apple access to their model and its backing dataset for a mere $1B/year unless there are other benefits that Apple will provide in return. Either Apple are paying several (ie, hundreds) of times more than what is being reported, or Apple are providing something else to Google (and we know Google is interested in only one thing).I don’t understand why people think this deal means that they’re just going to put Gemini on iPhones and call it Siri.
Totally this. It's quite unlike Apple to not buy an LLM that is already established and instead partner with Google. I guess in their head its the same as having Google as the default search engine? (which it totally isn't!). I have an iPhone 13 mini, so I probably won't get it anyway, but I hope there is an option to just use good old dumb Siri (who straight out blanks you if it can't help), rather than a Gooogle AI verison.• FFS Buy perplexity already‼️
What about Le Chat Mistral. As a european, I'd rather have an LLM hosted in europe to be honest, with the present political climate.How would that help? They don't make a model. They just use others and build tools around it.
What advanced AI models are included in my subscription? | Perplexity Help Center
This article lists the advanced AI models available on each Perplexity subscription, and explains what changes the list you see.www.perplexity.ai
Article says gemini powered revamped siri comes until end of this yearIt’s supposedly coming out Spring this year. Not sure where you got the info of it coming in Winter.