Apple WWDC on 10 June – what to expect

Apple has announced the date of this year’s World Wide Developers Conference (WWDC). It’ll take place from the 10 to 14 of June 2024. On day one, Apple will present new products at a keynote. Concerning hardware, the Mac Mini is likely to receive the M3 chip. An M3 Ultra Mac Studio is also conceivable. There are also rumours of new AirPods, although these traditionally tend to arrive in September along with iPhones.

But the star of WWDC 2024 will be software. In addition to new versions of its operating systems, Apple will be presenting its entry into generative artificial intelligence (AI). This hasn’t been confirmed quite yet, but in his X post on WWDC, Head of Marketing Greg Joswiak lets a spoiler slip. Apparently, the conference will be «Absolutely Incredible» – AI, get it? Artificial Intelligence?

The starting position

Apple is late in introducing its AI – Microsoft already has market leader OpenAI on board and is neck-and-neck with Google for the best chatbot. Meta’s own Large Language Model (LLM) has also been out for a while. Meanwhile, Apple’s Siri is still dumb as a post.

Time is slipping away. Apple needs to take action. Although the iPhone manufacturer’s often late to new trends, it implements them more thoughtfully. However, AI development’s currently progressing so rapidly that even Cupertino is coming under pressure. Unlike hardware, chatbots can only be developed to a limited extent in hermetically sealed laboratories. They need to learn, as only broad tests reveal weaknesses. The wild early days of ChatGPT and Bard have shown this very clearly.

Apple wants a slice of the pie

CEO Tim Cook knows this. Behind the scenes, AI development at Apple must’ve been running at full speed for a long time. Last summer, Bloomberg’s Mark Gurman reported on internal tests of an Apple GPT. As both customers and investors were becoming impatient, Cook went so far as to give an official confirmation in the last earnings call – Apple would introduce generative AI features this year.

This rarely happens. Normally, Apple lets rumours be rumours. People only talk about products when they’re ready. In addition to Tim Cook’s comment, there were other signs of Apple’s imminent entry into artificial intelligence. For example, the mere use of the term. Until recently, Apple consistently spoke only of machine learning. But suddenly, the new M3 MacBook Air is «perfect for AI». And when Apple scrapped its car project a month ago, most of the staff moved to the generative AI development department.

Apple will finally jump on the hype train at the World Wide Developers Conference (WWDC). The event revolves around software, and the unveiling of a new artificial intelligence seems a foregone conclusion.

Expected AI features

Exactly how and where Apple intends to use generative AI isn’t set in stone. A list of the rumours:

• Siri: the voice assistant gets a brain upgrade. It may not become a chatbot à la ChatGPT, but it’ll at least be significantly better than before.
• Apps: Apple’s native apps such as Music, Pages and Keynote will receive AI support. For example, to create playlists or summarise text – similar to Bing integration in Microsoft’s Office 365.
• Autocorrect: typing could become faster thanks to new AI features predicting words and sentences better.
• Image editing: Google and Samsung already offer various AI features for image editing. It seems inevitable that Apple will follow suit.
• Programming: end consumers are unlikely to know tools for developing apps, but they’re all the rage among developers. The Xcode programming environment is to receive comprehensive AI support.

In-house development or outside help?

There are currently growing indications that the generative AI behind these new features won’t come from Cupertino. According to the New York Times Apple is in negotiations with Alphabet. Talks are also underway with OpenAI. In the end, LLMs Gemini or ChatGPT could be under Siri’s hood soon.

The AI may even be different depending on the region. According to China Daily, Apple is looking into a collaboration with Baidu in China. However, an agreement hasn’t been reached yet.

These tactics don’t correspond to Apple’s normal procedures, with complete control over its own products and software. However, it wouldn’t be the first collaboration with other tech companies. Apple, for example, receives 18 billion dollars a year from Alphabet so Google remains the default search engine on iPhone – and Apple doesn’t develop its own.

It’s conceivable that generative AI will work in a similar way. Catching up to OpenAI and Google with its own creation would be an expensive Herculean task, so a partnership seems tempting. One interesting question is whether AI companies would have to pay Apple to gain access to users – or would Apple pay license fees for the technology?

Despite the negotiations, native Apple AI isn’t off the table yet. A research report suggests that the company’s still investing considerable resources in the development of their own LLM. It runs under the name MM1, which isn’t explained any further, but could stand for MultiModal 1. In other words, the model doesn’t only handle text, but also other media such as images. Just like Google’s Gemini.

What Apple could do better

Whether with its own model or an external one, Apple could differentiate itself from competitors such as Microsoft in various ways:

• Deeper integration: since Apple manufactures its own hardware, the AI can access more functions. This offers opportunities, but also potential for abuse. Imagine Siri could send money via Apple Pay by voice command – practical, but dangerous.
• Local processing: since the M3 generation, Apple’s been adding more AI capacity to its processors. The chip in the upcoming iPhone 16 will also be designed with this in mind. This would allow local processing, at least for simple requests. Complex calculations will probably still take place in the cloud.
• Data protection: Apple has long placed a focus on privacy. The company enjoys more trust than other large tech companies, an important factor in AI functions. After all, users disclose all kinds of data that may end up in a cloud.
• Curation: current chatbots often provide false information, sometimes without citing sources. Apple could do better. However, tighter guard rails would also limit capacities.
• Copyright: Reuters reports that Apple wants to make deals with media companies to obtain training data for an AI model. This would be in stark contrast to the strategy of OpenAI and Google, who train their models across the entire web without explicit permission. OpenAI’s already being sued by the New York Times.

We won’t know for sure which rumours are true until June. A lot of things will likely be decided at the last minute, which is atypical for Apple. It’ll be interesting to see whether Tim Cook can fulfil the high expectations.