Apple’s Plan to Win the Generative AI Race Involves Weird Emoji

Apple, the company that brought you those animated Memoji you never use on the iPhone and made Personas so you can video chat as a character straight out of Polar Express while wearing a Vision Pro, has a plan to win the generative AI race: weird emoji.

While companies like ChatGPT, Google and Microsoft have a head start using artificial intelligence trained on large language models to generate text and conversations that sound more human, remove items and people from photos, and develop tools to make weather forecasts more accurate, Apple is, according to a Bloomberg report, going to offer the ability to put AI-generated emoji in your chats.

And people will love it — because this is Apple. 

Of course there is much more to Apple’s future AI plans than just the ability to make any emoji you can imagine (or rather not imagine) a reality. 

At WWDC, Apple’s big developers conference that starts on June 10, the iPhone-maker is expected to show off new versions of its operating systems like iOS 18 and debut generative AI tools. When I first heard about Apple’s reported AI features, I was fascinated, and not just by the food-animal hybrid emoji possibilities. I was genuinely filled with hope that Apple could actually set itself apart from all the other AI criticism and controversy currently happening in tech.

But let’s back up a bit to set the scene.

Back in February, during a first-quarter earnings call, Apple CEO Tim Cook told shareholders that the company is putting “a tremendous amount of time and effort” into integrating AI into its products. Then in May, on a second-quarter earnings call, he said that Apple was making a significant investment into generative AI. 

While we expect to hear a lot about AI at WWDC, Bloomberg reporter Mark Gurman got the tech circles buzzing with his report on what AI features could be coming to iOS 18 and MacOS 15.

We could see AI help edit photos, send messages or reply to email. AI could summarize long messages and transcribe voice messages. Search could be smarter, and Siri could speak more conversationally. AI tools might even summarize something you come across browsing on Safari. A lot of these features wouldn’t be that surprising because they already exist on the latest Samsung Galaxy phones and Google Pixels.

But there are three possible Apple AI tools in Gurman’s report that stood out to me.

On-the-fly AI emoji, please and thank you

While there already exist tools and sites that can mash up emoji icons into ridiculous symbols, like Emoji Kitchen (I’ve talked about the topic before on CNET’s One More Thing show), Apple might launch its own version of this tool in iOS 18. 

Bloomberg says that the iPhone might be able to create an emoji for you on the fly, something outside the standard emoji library set by the unicode consortium. Apple knows how to have some fun with messaging, and it’s right out of the company’s playbook to give folks something silly to talk about in a software update. 

When Apple does a weird cartoony thing, that’s the feature that gets all the attention. When the Vision Pro came out, we all gawked at the Personas. The iPhone X’s true depth front-facing camera was fun to show off because of the animated Memoji faces in chat. And when Apple announced the iPhone 11, it was the ability to record slow-motion videos with the front-facing camera that made the phone more exciting (I’m glad the term “slofie” didn’t catch on — nice try, 2019 Apple). 

But I am being serious when I say that Apple is really smart to give people a lighthearted playground for AI image generation. We are in a time filled with stories of how AI systems are ethically questionable and making terrible goofs.

Recently, Google set its Gemini AI free on search, and folks poked fun at the weird results it gave. Gemini pulled jokes from Reddit and The Onion, then spewed out answers as hard facts, telling people that eating rocks is good for you and recommending people put glue on pizza. 

That same week, movie star Scarlett Johansson publicly criticized OpenAI and got her lawyers involved over its lack of transparency. ChatGPT’s flirtatious chatbot voice sounded eerily like her, despite Johansson denying the company permission to use her voice. The voice has since been removed. 

With emoji creation, Apple’s implementation could get customers interested in using AI while avoiding any controversy. This is assuming that people don’t make something terrible with the emojis. Let’s not go crazy with the eggplants, people. 

Siri reboot: Now with more AI

One thing I’m watching out for at WWDC is if Apple’s personal assistant, Siri, gets an upgrade to better understand our commands. The Bloomberg report mentioned that a more advanced Siri would be coming to not just the iPhone, but also the Apple Watch for on-the-go tasks. 

A revamped Siri that works well on the watch is exactly how Apple can stand apart from Google, Microsoft and OpenAI. If Apple has a useful AI helper on a wearable, it would do what other companies have been trying and failing to do so far: Make an AI companion that can go anywhere without needing the phone.

Read more: AI Could Be the Key to Making Smartwatches Better

A month ago we saw the launch of the Humane AI Pin, a $700 AI computer you wear on a shirt. But early reviews blasted the wearable for being buggy and not fulfilling its promises. Humane, the startup behind the AI Pin, is now reportedly trying to sell itself for as much as $1 billion.

Last month was also the launch of another tiny AI-powered handheld gadget called the Rabbit R1. It also tried to be an AI assistant that you could take anywhere and answer questions, but it didn’t answer much and left tech reviewers disappointed. 

If Apple brings a more AI-savvy Siri to the Apple Watch, then it will show all these other companies how a real AI voice assistant can actually be helpful on the go, instead of just being hyped as helpful. 

But that feels like a big “if.”

Other companies are already dabbling with putting more AI on watches as a health or workout assistant. The Amazfit Cheetah has an AI running coach of sorts. Google’s FitBit Labs is launching a new program that uses generative AI to answer questions and provide more context about your activity data. The idea is that the AI could be used to help you set realistic fitness goals.

A year ago, Bloomberg reported that Apple was working on some sort of AI-powered health coaching service. But who knows how close we are to seeing that launch. Apple also took some baby steps by putting more powerful chips in the latest watch models. That means Siri requests are processed on the device faster and more securely, instead of having the data be processed in the cloud. The Apple Watch Series 9 and Ultra 2 let you ask Siri about your health data or log information about your health. And it’s all done locally on the device.

Surprise twist: More privacy

Another area where Apple could stand apart from everyone else is privacy. It’s like that old ad, what happens on your iPhone stays on your iPhone. 

Bloomberg reports that much of the processing for Apple’s AI in the next version of iOS and MacOS would run entirely on the device. 

But there’s a twist that worries me.

It’s also being reported that if an AI feature requires more horsepower, the work will be pushed for processing in the cloud, to data centers powered by Apple’s M2 Ultra chips. 

So Apple will really need to explain how our data is being protected and give examples of what tasks might need to be calculated in the cloud. What is being sacrificed here? The Information reports that Apple might protect cloud data by using some sort of virtual black box.

If Apple’s AI system can summarize our private notifications, our text messages, the sites we visit and the media we consume, Apple needs to be clear about how it’s handling our privacy. 

There are already many questions on what data these AI companies are slurping up to train their systems, and what we are giving away to feed these beasts. So Apple needs to be prepared to talk about privacy, which might be harder to do because of these last two rumored announcements.

The same reports say Apple will announce it partnered with OpenAI somehow. I don’t know how Apple is going to walk this line of promoting a safe, private environment, but also boasting that it invited a troublemaker into the backyard to play in our sandbox. Even if you trust Apple, are you ready to trust Apple’s friends? 

Apple’s AI already caused drama before it’s announced

Another concern comes from a rumored feature on the Safari browser called Web Eraser. As reported by Apple Insider, the tool lets you take out any element of a website, like an ad. And Business Insider reports that a group of French publishers and advertisers are asking Tim Cook to halt deploying this tool because of how it could hurt publishers that rely on advertising for their business. It’s not even out yet and there’s already a signed letter against it, so we’ll have to see what that’s all about. 

So how does Apple dance around this drama? Easy. Tell everyone it’s just in beta. That’s what Bloomberg is reporting. Apple will likely label AI features as beta or that they’re still being tested. 

Oh sure, every single iPhone customer will try it. But hey, you can’t get mad at Apple if it hangs out an “under construction” sign and its AI becomes wonky and falls apart on you, misses a fact in the summary, skipped a word in a transcription, or causes your website to fail because you can’t make money anymore.

It’s like being confidently unconfident.

But even with a beta label, there’s one slipup that people won’t forgive: a privacy slipup.

CNET’s Patrick Holland edited this report.

Editors’ note: CNET used an AI engine to help create several dozen stories, which are labeled accordingly. The note you’re reading is attached to articles that deal substantively with the topic of AI but are created entirely by our expert editors and writers. For more, see our AI policy.