Apple Tests 'Apple GPT,' Develops Generative AI Tools To Catch OpenAI (bloomberg.com) 32
Apple is quietly working on artificial intelligence tools that could challenge those of OpenAI, Alphabet's Google and others, but the company has yet to devise a clear strategy for releasing the technology to consumers. From a report: The iPhone maker has built its own framework to create large language models -- the AI-based systems at the heart of new offerings like ChatGPT and Google's Bard -- according to people with knowledge of the efforts. With that foundation, known as "Ajax," Apple also has created a chatbot service that some engineers call "Apple GPT."
In recent months, the AI push has become a major effort for Apple, with several teams collaborating on the project, said the people, who asked not to be identified because the matter is private. The work includes trying to address potential privacy concerns related to the technology. [...] Apple employees say the company's tool essentially replicates Bard, ChatGPT and Bing AI, and doesn't include any novel features or technology. The system is accessible as a web application and has a stripped-down design not meant for public consumption. As such, Apple has no current plans to release it to consumers, though it is actively working to improve its underlying models.
In recent months, the AI push has become a major effort for Apple, with several teams collaborating on the project, said the people, who asked not to be identified because the matter is private. The work includes trying to address potential privacy concerns related to the technology. [...] Apple employees say the company's tool essentially replicates Bard, ChatGPT and Bing AI, and doesn't include any novel features or technology. The system is accessible as a web application and has a stripped-down design not meant for public consumption. As such, Apple has no current plans to release it to consumers, though it is actively working to improve its underlying models.
Software software (Score:1)
everyone is working on software .. how about some hardware .. robots specifically. Robotics hardware sucks. Everything about robotics hardware is subpar. Robots walk with a awkward gait, consume mad energy, and their hands/grippers suck ass .. they can't do simple things like pickup a rubber band or manipulate a small screw.
Re: (Score:2)
X Company launches X-AI (Score:4, Interesting)
McDonalds tests McGPT, Develops Generative tools to Grill OpenAI
Nabisco tests OreoAI, Develops Generative tools to Twist, Link, and Dunk OpenAI
Pampers tests DiaperAI, Develops Generative tools to poop all over OpenAI
Kudos to Sam Altman for his dramatic reveal of OpenAI and LLMs, but seriously I've never seen technology fad develop so quickly and spread so rapidly. And yet the reality is LLMs are good for certain things and not for others; it's a powerful and complex tool but it's also plagued with problems that are intrinsic to the nature of LLMs, such as the amount of manual labor to create the metadata that allows LLMs to work [theverge.com] which is a direct source of problems of bias [cbsnews.com] and fake references [duke.edu] for example.
ChatGPT and other LLMs are here to stay, but it just feels like the field is rushing like lemmings trying to jump on it because everyone else is without really thinking about what it's useful for. This looks to me like a bubble in the making and what will be a lot of burned money and value and wasted effort until it all burns down and what's left is what it's actually good for.
x-AI does not belong in that group. (Score:1)
A good summary but note that x-AI is not at all about LLMs, but AGI (artificial general intelligence) looking to answer much harder questions.
Not going to get there from here (Score:1)
Generative AI is interesting and potentially useful - ala a self-training expert system - but AGI is not coming from this direction. It's a better Eliza overlaid with voice recognition. More capable of passing a Turing test than Eliza was, but as Eliza wasn't anything like a real intelligence, this is also subterfuge.
Re: (Score:2)
Re: (Score:2)
General artificial intelligence (or whatever you want to label human intelligence today) will not come from computers the way we make them today (CPU/GPU) because those programs are deterministic themselves in nature
Determinism has nothing to do with intelligence.
Re: (Score:3, Informative)
All neural networks are pre deterministic in nature.
Many argue that so is the human brain, that there is no free will.
Re: (Score:3)
"Synthetic Intelligence" would be a good name, if we ever achieve such a thing.
Or possibly "Machine Intelligence" to clearly distinguish it from "human intelligence" specifically, or "organic intelligence" more generally.
Incidentally, there is NO component to human thought that is both non-deterministic and non-random. Anything that might seem to be in the middle is just state-based, with "state" being composed of quite a large list of unknown variables.
Re: (Score:3)
ChatGPT and other LLMs are here to stay, but it just feels like the field is rushing like lemmings trying to jump on it because everyone else is without really thinking about what it's useful for. This looks to me like a bubble in the making and what will be a lot of burned money and value and wasted effort until it all burns down and what's left is what it's actually good for.
LLMs could be a bubble, but companies were burnt 10 years ago for thinking the same about DNNs. The potential blockbuster potential against the relatively minor investment in LLMs (for a large company) far outweighs any possibility of a bubble. It's a no-brainer to jump on this bandwagon (for large companies).
Apple has around $200 billion in cash. It would be very stupid of them not to spend a small fraction of one percent of their cash to pursue LLMs, especially if this direction has any possibility of
Re: (Score:3)
ChatGPT got prominence for one reason: it gave seemingly better results than Google, and as a result it made Google which has always seemed unassailable now potentially vulnerable. Google had a knee-jerk reaction to it. But Apple is methodical and analytical; this is very quick for them to investigate it. It just feels like an enormous fad still.
Re: (Score:2)
I don't know if that's worth the millions they'll probably end up investing in this, but I suppose that I'd like it a lot more than most of the crap they put in their more recent OS releases.
Re: (Score:3)
> I've never seen technology fad develop so quickly
> and spread so rapidly
Did you sleep through the while blockchain/crypto/NFT fad? Yeah... at this point, people have figured out that it's a scam and the previously intolerable cacophony has dulled down to a merely obnoxious vuvuzela drone. But for a long while there, my LinkedIn was getting crapflooded by multiple fly-by-night crypto startups daily... more than a few of which wanted to pay me in whatever their bitcoin knock-off was instead of real
Re: (Score:2)
Blockchain LLM is the future. It lets you trace whatever hallucination the AI comes up with all the way back to the source training material. It only uses a few GW more power than competing LLM options.
Re: X Company launches X-AI (Score:3)
Bitcoin is a fad that after being busted, is worth only 500 billion dollars.
Re: (Score:3)
Another perspective:
Apple launches X product to try to catch up with market leader.
Other examples:
- Apple Maps (Google)
- Apple TV (Netflix)
- Vision Pro (Oculus)
- Apple watch (FitBit)
- Apple Car (Waymo)
Re: X Company launches X-AI (Score:2)
Are these just all skins around ChatGPT though? Doing API calls to OpenAI? Lol.
Re: (Score:2)
Old MacDonald had farm AI,
He called it AI-AI-O.
Siri (Score:2)
Apple, being a consumer products manufacturer, will focus on the utility and not the âoemagicâ. Siri will be smarter and perform a lot more functions that your computer can already do, but we donâ(TM)t have an easy way to ask.
Re: (Score:2)
Re: (Score:2)
As long as they are smart and they constrain its creative output. Use the LLM to interpret intent and then rely on traditional sources to respond. It would still be way better than it is now. Letting it hallucinate "facts" is probably not going to be a killer product.
Re: (Score:1)
Compete with OpenAI not catch OpenAI (Score:2)
Referring a service as being "to catch OpenAI" suggests it's a tool to detect whether a piece of text came from OpenAI's tools. Which would, be very useful, by the way. At this rate All the search engines are going to need ChatGPT detectors, Otherwise all our search results are going to be full of Spam reponses from SEO companies who use GPT to create Fake blogs for the purpose of attracting search engine and social media hits on keywords To then refer to their real websites (For rank boosting
Re: (Score:2)
Flagging ChatGPT might be easier than the others by just looking for the boilerplate style from its guardrails but it might still have more false positives than real ones. Maybe nobody will automatically publish "As a large language model"
Not Surprising (Score:3)
Every large tech company is doing the same. It's not surprising. Apple's approach will likely be a bit different, with their push for privacy and will likely rely more on keeping the processing on-device.
It'd be more surprising to learn a tech company ISN'T exploring their own AI.
Re: (Score:2)
It's too bad all the companies that offered a voice assistant already cried wolf before even having a decent chat bot. Everyone got tired of trying to get Eliza to be useful.
I'd be surprised if on-device is feasible yet, though, because cutting the RAM really cuts down on how much it can do.
Ajax (Score:2)
Good, cause it's not like the term "Ajax" has ever meant anything else for developers or the world at large
Re: (Score:2)
They have to appear at the beginning of the phone book. A1 looks too much like AI and nobody would let that go.
Are you getting Newsletter popup notices? (Score:2)
Translucent Plastic (Score:2)
still not sticking to Apple AI.
Brains for Siri (Score:2)
Would be nice.
thank god. (Score:2)
cuz ffs, siri.