Google and Microsoft have made their developer conferences a showcase of their generative AI chops, and now all eyes are on next week’s Worldwide Developers Conference, which is expected to mark the debut of Apple Intelligence.
The Cupertino-based company is facing a lot of pressure. Apple has fallen behind its peers in the AI race, and it probably feels like it needs to pull out all the stops to impress fans and shareholders. But that shouldn’t mean overpromising on features.
Reliability first
Apple makes some of the most popular devices on the planet, and its AI features should serve to make them more useful. A lot of AI-powered features rely on going back to the cloud to get answers or inputs back. However, if Apple manages to run some useful features locally on-device, users might ditch the cloud-based tools in favor of always available AI. Offline transcriptions in the Voice Memo and Notes apps could fit the bill.
Apple will likely reveal summaries of notifications and web pages, basic text generation and photo editing. However, tons of browsers, note-taking apps, and photo-editing apps already have those. Apple needs to make its implementation as smooth and seamless as possible to make it stand out.
Privacy first
Apple is likely to bolster its privacy-first approach, so it might not give Siri or AI-powered features free rein to take control of all apps. According to a Bloomberg report, only the iPhone 15 Pro and iPads or Macs with M1 or later chips will get AI features, and they will be opt-in. If this is true, despite lagging the AI feature adoption curve, Apple is still cautious and doesn’t want to be caught in the user backlash.
The company was recently criticized for its iPad “Crush” ad, which showed creative instruments being destroyed under a hydraulic press. This was seen as Apple undervaluing creators, their tools, and the effort it takes to make art by packaging it into a slim capitalist package. With AI already having a bad rep among creators, Apple might not want to irk them again. So it will likely take a non-controversial approach.
Improving Siri
The biggest change expected is for Apple to revamp Siri to understand users’ queries better and deliver more accurate results. Currently, Siri can’t multitask. If you ask the assistant to set a 10-minute timer and a 5-minute timer, it will set one for 15 minutes instead. These things might not need generative AI’s help to solve, but Siri’s revamp should at least have them.
If Siri doesn’t get deeper access to apps as expected, Apple can make users’ lives easier by introducing an AI assistant to help users create complex Siri shortcuts to achieve multistep tasks.
Rumors on the street are that Apple will announce a deal with OpenAI to power AI features across its operating systems. It remains to be seen how much of Apple Intelligence will be built on that deal. Given AI’s hallucination problems, Apple might not want to be directly involved in content-related AI features just yet.
Lots of companies make big promises about AI-powered features, only to disappoint with inaccurate or biased results. Companies like Google and OpenAI have had to walk back on AI features because of errors or copyright issues. As such, the company may not want to rely on LLMs (large language models) for content generation.
