The hype machine is real with Generative AI and ChatGPT, which are seemingly everywhere in tech these days. So it's no surprise that we're starting to hear about a new and improved Siri. In fact, 9to5Mac has already detected a new natural language system.
You speak my language?
The claim is that Siri in tvOS 16.4 beta has a new “Siri Natural Language Generation” framework. As described, it doesn't sound impressive, as it seems mostly focused on telling jokes (to dad?), but it can also allow you to use natural language to set timers. His code name is "Bobcat".
These whispers follow a recent New York Times report on Apple's artificial intelligence summit in February. This report claimed that the event emphasized the type of generative content and extensive language models (LLM) used by ChatGPT. He also said that Apple engineers are "actively testing" language generation concepts by releasing new language concepts every week as Apple seeks to advance AI.
So are you creating a competitor to ChatGPT? Not really, according to Bloomberg.
"Hey Siri, how do you spell 'catch up'?"
While Siri seemed incredibly sophisticated when it first appeared, development hasn't kept pace, giving Apple's cheeky voice assistant echoes of MobileMe and Ping. Like Apple's two flops, Siri promised it never quite lived up to it and now lags behind Google and Amazon Assistants, albeit a bit more private.
Siri's lack of contextual sense means it's only really good for what it's trained to do, limiting its abilities; GPT seems to leave it in the dust. With the recent update to GPT-4, OpenAI is innovating rapidly. We can already see that this has lit a fire at big tech companies. Microsoft has embraced ChatGPT on Bing, Google is making rapid progress on Palm development, and Amazon is pushing hard on AWS Chat (the latter now integrated with Microsoft Teams).
Apple, and Siri, seem to be on a limb.
It's not the only one
Of course, Siri isn't the only artificial intelligence (AI) Apple is working on. In some areas, like accessibility and image magnification, it achieved incredibly good examples of instant messaging done right. But somehow, Siri always makes mistakes.
I'm not sure how Apple's Steve Jobs would have handled this - he's not happy when his HomePod tells him it can't find his Dylan songs. The difference between the two voice-capable AIs is that he could have GPT create an image of him throwing that smart speaker at the wall.
This is partly due to the way Siri is built.
How Siri was created
Siri is something like a huge database of answers for different knowledge areas supplemented with Spotlight search results and natural language interpretation so you can speak to it. When a request is made, Siri verifies that you understand the question and then uses deep/machine learning algorithms to identify the appropriate response. To get this answer, you perform a numerical assessment (confidence score) of the probability that you have the correct answer.
Esto signifies that cuando le haces una pregunta a Siri, first echa un vistazo rápido para ver si se trata de una simple solicitud ("encender las luces") a la que puede responder rápidamente à partir de lo que ya sabe, o si debe consultar at the más grande. date base. Then it does what you tell it to do (sometimes), gets you the data you need (often) or tells you it doesn't understand you or asks you to change a hidden setting somewhere on your system ( demasiado a menudo).
In theory, Siri is only as good as its database, which means the more responses it contains, the better and more efficient it is.
However, there is a problem. As former Apple engineer John Burkey explained, the way Siri is built means engineers have to rebuild the entire database to update it. This is a process that can take up to six weeks.
This lack of actual learning makes Siri and other voice assistants "dumb as a rock," according to Microsoft CEO Satya Nadella. Of course, I would expect it to say something like that, because Microsoft has invested billions in ChatGPT, which it integrates into its products.
Generative AI, on the other hand.
Generative AI (the kind of intelligence used in ChatGPT, Midjourney, Dall-E, and Stable Diffusion) also uses natural language, its own databases, and search results, but it can also use algorithms to create original content like audio, images or text.
You can ask it a question and it will review all the available data and make some decisions to get a result.
Now, as has been pointed out quite often since people began exploring the technology, these results aren't always great or original, but they generally seem convincing. The ability to ask you to generate fake videos and photos goes one step further.
In use, one way to see the difference between the two AI models is to think about what they can accomplish.
So whereas with Siri you can request a map of Lisbon, Portugal, or even find directions to somewhere on that map, the generative AI allows you to ask more nuanced questions, like parts of the city you recommend, to write a story. with the action based in that city, or even create a creepy fake photo of you sitting in that cute bar in Largo dos Trigueiros.
It's pretty clear which AI is the most impressive.
What happens next?
It does not have to be this way. Developers have successfully created apps to add ChatGPT to Apple products. watchGPT, which was recently renamed Petey – AI Assistant for branding reasons, is a prime example.
Apple is unlikely to want to sell such a competitively important technology to third parties, so it's likely to continue working to find its own solution, but that could take years, during which time Siri might still not open the app. cabin door.
However, since GPT-4 costs as much as 12 cents per thousand requests, it is highly unlikely that Apple will integrate it directly into its operating systems. With an installed base of over a billion users, that would be extremely expensive, and Microsoft is already there.
It's in this context that Apple could simply bite the bullet to make it easy for its developers to add support for OpenAI technology into the apps they create, effectively passing the cost on to them and their customers.
It might help in the short term, but I'm sure it's a fire in the stomach for Apple's AI teams. Now they will be doubly determined to scale the innovation in natural language processing that lies at the heart of both technologies.
But at this point, in terms of implementation, they seem to have fallen behind. Although appearances, as shown by the images generated by GPT, can be deceiving.
Follow me on Mastodon or join me at AppleHolic's bar & grill and Apple discussion groups on MeWe.
Copyright © 2023 IDG Communications, Inc.