It’s Apple v Microsoft as tech titans battle for AI supremacy

Though generative AI has developed at a breakneck speed over the past few years, it is only now being meaningfully integrated into consumer devices. That means technology giants – specifically the ones making those devices – are moving to present their own takes on AI as the best customer experience.

Last week, Apple unveiled Apple Intelligence, which joins Microsoft’s Copilot as a brand-specific AI that makes use of a new generation of neural processors to do things such as generate images, rewrite text and transcribe or translate language in real time, without needing to connect to the internet. However, each also has specific features to tempt you into a specific ecosystem and keep you there. Here’s how they are shaping up.

Siri and Apple Intelligence can use Chat-GPT for some requests, such as generating or refining text.

Apple goes private and personal

Apple is aiming for ubiquitous generative AI tools across all its devices, but with the updates only coming to the Pro versions of the iPhone 15 – and presumably all models of this year’s iPhone 16 – the first place many people will experience them is on the Mac. Expected to release in beta in September, these tools will work on any Mac with an Apple Silicon chip, which is most models released in the past four years.

A new Writing Tools panel can show up essentially anywhere you enter text. There are quick options to make your writing more friendly, more professional or more concise, or you can describe the change you want to make.

Smart Reply will boil a message you receive down to a series of questions, and generate a response.

You can also summarise any text, or reformat it into a list, table or bullet points. Apple has also shown off an expanded Smart Reply ability that analyses a message, asks you a series of questions and then formulates a reply.

Image generation has generally been a bit tougher for which to find a mainstream use, but Apple is experimenting with a few different forms of it.

A new Image Playground app lets you generate images by describing them or selecting themes, and you can even point to a person from your photos to make an image that looks like them. However, the results appear consistently “cartoony” rather than realistic.

Playgrounds is integrated into other apps where you might want to whip up a custom image as well, and anywhere you find emojis you can also create custom icons by describing them.

Taking a cue from some of its rivals, Apple will also be able to scan your photographs and remove unwanted background elements, or turn your sketches into AI illustrations, with a tap.

With its revamped Siri, Apple seems to want its AI to be more personal than the creativity focused chatbots we have seen, securely soaking up context from all of a user’s data and behaviour in the background, so when you ask for something it knows what you mean. For example, if you mention something to do with “the recipe from yesterday”, it should know what you are talking about.

Though we are still a few months away from Apple Intelligence, the announcements were met with some trepidation. Most AI processes will be handled on-device by Apple’s processors, but some will need to run through cloud servers to make use of larger and more complex language models.

Apple says it has created secure private servers that won’t store information, but has not clarified if users will know when their information is moving from their device, or if they will be able to turn it off.

To provide the broader chatbot experience some users want, Apple is partnering with other AI providers, much in the same way it partners with search engines rather than develop its own. So far, it has only announced compatibility with OpenAI’s ChatGPT, which can accomplish tasks beyond Apple Intelligence, such as drafting entire scripts. Apple has been clear that no personal data from this process is stored by OpenAI, but the announcement still sparked some privacy concerns, and also led some experts to worry about the amount of data being centralised at OpenAI that it could use to further develop its products.

Users who pay for an OpenAI subscription will be able to access more powerful capabilities directly through their Apple devices.

Microsoft Copilot+ and the ARM race

Microsoft is a major investor in OpenAI, but rather than use ChatGPT directly it has adapted the technology into its own offerings, including the AI-powered search engine Bing and various Copilot features that show up in everything from Office to Teams. However, its big consumer push is to integrate Copilot into the Windows and Surface experience, as it moves to head off Apple and its AI-enhanced MacBooks. In fact, you could argue that AI is bringing us back to the good old days of PC vs Mac.

Apple had differentiated itself by creating its own ARM architecture Apple Silicon chips, leaving Intel’s x86 processors behind. But now, in its pursuit of AI power and Mac-beating efficiency, Microsoft is doing the same, with ARM chips from Qualcomm featured in its latest Surface portable PCs due for release this week. It has created a new designation for Windows machines with powerful AI-capable chips and its full Copilot experience (for which other device-makers using ARM or x86 chips can qualify) called Copilot+ PCs.

Recall, which will work Copilot+ PCs, remembers everything you do on the PC so you can scroll back through time.

These devices have powerful neural processors that allow on-device AI to run much more efficiently than they would on other Windows PCs, and they have a dedicated key to summon the CoPilot chatbot which, like Apple’s Siri, will be able to take context from whatever you are doing at the time you are asking questions.

Like Apple, Microsoft says some AI experiences will need to be processed in the cloud and will be done securely, thanks to a Microsoft-designed security processor in every CoPilot+ PC, but it is unclear where that line between local and cloud is. However, Microsoft is focused on promoting a few AI tasks that are completely local and exclusive to Copilot+.

One is Recall, which keeps track of everything displayed on your screen to help you remember it later. Microsoft said it has designed this to be a more friendly alternative to trying to remember what a file was called, or what website had the thing you were reading, or discerning what a picture was when it is called IMG3114.jpg. With Recall, if you were trying to find that video of a dog you were watching earlier, you would just type “dog” and it would take you back to the Reddit post, even if the word dog didn’t appear anywhere on the page.

Microsoft’s Cocreator draws alongside you to create an AI-generated image.

Another Copilot+ feature is Cocreator, which lives in apps such as Paint. You tell it what you are attempting to draw, and the AI interprets your drawing in a separate window. There’s a “creativity” dial you can use to customise that interpretation, giving you more control over composition than standard AI-generated artwork.

Of course, hitting the market with an AI PC first also opens Microsoft up to more scrutiny, which we have already seen with Recall. The company seemed unprepared for the backlash against this feature, with security experts wondering if a stolen PC could hand full histories of passwords, banking information and black-mailable material to thieves.

The company clarified that this was an entirely on-device feature that was securely encrypted and accessible only to the person enrolled in the device’s password or biometric locks. However, ultimately, it walked the feature back and said it will be “off” by default, and will require biometric authentication and proof of presence to decrypt.

Most recently, Microsoft delayed the feature, saying it would launch next week only for users enrolled in its early-access Windows Insider program, before launching for everyone some time in the future.

Get news and reviews on technology, gadgets and gaming in our Technology newsletter every Friday. Sign up here.