Powerful LLM, you mean. AGI isn't there, and with this approach won't be - MS may have all the models, but if it still misses people who are needed to make better models, it's a delusion in my book. Let Satya think he got it covered.
Who could be in command of a powerful AGI? Couldn't anyone with temporary access to the system have it generate a game plan for how that person can steal the agi?
That is very likely a longer term goal for Apple, but they want something ready to go this year and not to have a Google Gemini type half-baked launch.
They almost assuredly have and will continue to. But it may not be ready fast enough.
Putting an AI assistant in every Apple product (it would have to be ecosystem wide) is nowhere near the same use case as dropping a model like llama on the web and saying it’s for research only.
Apple has successfully leaned on other company’s services for core features (Google maps) and even then it took years and years to create a reasonably compelling first party, privacy preserving alternative.
The Vision Pro is the company’s most risky bet in some time releasing a product that is not quite there. Reporting is there was great rancor over whether to continue to wait.
The company is exceedingly conservative about major changes to how its consumer-facing products behave and releasing anything new that isn’t ready.
This is true inclusive of the VP (which I think was well baked), AirPower (which clearly was not) the original iCloud and a handful of other stand out miss-calculations and failures.
A deal with OpenAI provides leading edge expectations of today’s consumers for AI assist. So, they won’t lose ground while Apple works quietly in the background.
Instead the company can watch behavior, develop their idea for how it aught to be and build in the background on their own timeline.
It's a good approach given the company was caught as flat footed as everyone else.
I can't imagine Apple would make a deal that would allow a Third Party company to use Apple users' data for training. I would bet they'll either put it on-device, arrange the terms so that OpenAI cannot train on user data, or a mixture of both.
They've kept this company propped up with billions in data-center expenses, fed it pipelines galore of data, and now johnny-come-lately shows up & reaps all the same reward. While not firehosing all their data back in.
Zero loyalty, zero benefit for Microsoft if this alliance they've spent out the ass for again and again starts partnering up with the direct competitor, who unlike them will retain all the data. But that's OpenAI!
Any company so brazenly willing to reneg on their not-for-profit mission, willing to completely two-facedly change themselves at a whim to chase profit: that should have been a sign. You should have known. Should have seen this coming.
At this point the company from business - not product - perspective looks like a hopeless cutthroating environment, when first they didn't let the failed exec go, then, too late, realized their mistake and started loudly complaining. If OpenAI will successfully survive, it deserves textbook mentioning in a chapter of business crises.
To give to context into this (with friends at Azure and competitors), MS has been heavily pushing LLMs for Enterprise SaaS usecases like Code Completion, Automated SOC, Auto-Remediation, Automated Asset Labeling, etc. It's been a major factor in driving Azure sales recently due to Azure OpenAI Services and multiple other features that are a reskin around OpenAI products.
An Apple dedicated deal means MS roadmap items will become much more tenuous, as MS's product priorities and timelines will begin clashing with Apple's.
Already MS competitors like Databricks, Snowflake, Crowdstrike, SentinelOne, Google, etc are all releasing product lines that touch on these usecases and are doing so by building their own in-house models.
MS's biggest moat in the Enterprise space recently has been it's partnership w/ OpenAI, which forced all the other competitors to spend 2-3 years building in-house models and teams.
Feeding frenzy. Who actually wants this? I want to use LLMs as anonymously as possible. Honestly, I find no use for them at all other than helping me plan travel itineraries.
22 comments
[ 3.2 ms ] story [ 55.0 ms ] thread"We are below them, above them, around them. " -Satya on OpenAI
/s
Putting an AI assistant in every Apple product (it would have to be ecosystem wide) is nowhere near the same use case as dropping a model like llama on the web and saying it’s for research only.
Apple has successfully leaned on other company’s services for core features (Google maps) and even then it took years and years to create a reasonably compelling first party, privacy preserving alternative.
The Vision Pro is the company’s most risky bet in some time releasing a product that is not quite there. Reporting is there was great rancor over whether to continue to wait.
The company is exceedingly conservative about major changes to how its consumer-facing products behave and releasing anything new that isn’t ready.
This is true inclusive of the VP (which I think was well baked), AirPower (which clearly was not) the original iCloud and a handful of other stand out miss-calculations and failures.
A deal with OpenAI provides leading edge expectations of today’s consumers for AI assist. So, they won’t lose ground while Apple works quietly in the background.
Instead the company can watch behavior, develop their idea for how it aught to be and build in the background on their own timeline.
It's a good approach given the company was caught as flat footed as everyone else.
Further so to see if it would be Nvidia GPUs (likely) or custom Apple Silicon (highly unlikely).
They've kept this company propped up with billions in data-center expenses, fed it pipelines galore of data, and now johnny-come-lately shows up & reaps all the same reward. While not firehosing all their data back in.
Zero loyalty, zero benefit for Microsoft if this alliance they've spent out the ass for again and again starts partnering up with the direct competitor, who unlike them will retain all the data. But that's OpenAI!
Any company so brazenly willing to reneg on their not-for-profit mission, willing to completely two-facedly change themselves at a whim to chase profit: that should have been a sign. You should have known. Should have seen this coming.
An Apple dedicated deal means MS roadmap items will become much more tenuous, as MS's product priorities and timelines will begin clashing with Apple's.
Already MS competitors like Databricks, Snowflake, Crowdstrike, SentinelOne, Google, etc are all releasing product lines that touch on these usecases and are doing so by building their own in-house models.
MS's biggest moat in the Enterprise space recently has been it's partnership w/ OpenAI, which forced all the other competitors to spend 2-3 years building in-house models and teams.