Apple’s belated AI gambit

AI supporters and sceptics alike have been looking for the technology’s watershed moment — the development that solidifies the usefulness of large language models to businesses or consumers and finally puts some heft behind surging AI valuations. Enterprise AIs have been rolled out, but it remains unclear how the technology will lift bottom lines. Consumer applications have been experimental and limited.

To many, the most fitting consumer use would be as a personal assistant embedded in a smartphone. So it was only natural that Apple, maker of the iPhone and its “digital assistant” Siri, announced its own AI personal assistant this week. Other AI players are heading in that direction, too. Google unveiled its AI assistant Astra last month, and OpenAI has been focused on making its model more consumer-friendly.

It is tempting to say that latecomer Apple’s entry into the AI race is the long-sought turning point. The tech behemoth’s large share of the smartphone market, its loyal buyers and its strong inter-device networks could make the company the logical consumer leader. Its strengths might be enough to usher in the age of AI transformation, after the current period of AI hype. 

Yet Apple has long been seen as the AI laggard. Lacking a well-defined AI strategy or publicly available model of its own, the company was overtaken in stock market valuation first by rival Microsoft and in recent days by chipmaker Nvidia. And many AI aficionados balked at Tim Cook’s modest computational goals for the company’s AI unit Apple Intelligence, arguing that rivals Google, Microsoft, OpenAI and Amazon are better positioned thanks to larger LLM models.

But Apple’s critics forget that ever since Steve Jobs’ passing, the company’s strength has been implementation, not innovation. It did not invent the tablet or smartwatch, but its large consumer base and focus on sleek design brought them into vogue. And the use of a smaller model in Apple Intelligence arguably better fits the needs of consumers. Most users will not need their AI assistant to compose stories in the style of The Arabian Nights; they will just need it to write emails. Apple’s partnership with OpenAI addresses concerns about a lack of computational power too, with users able to deploy ChatGPT to carry out more difficult tasks.

Whether this is truly a landmark moment, however, depends on the rollout. Shaky launches of Google’s now defunct Bard and its more recent Gemini tools have suppressed uptake. Microsoft’s spotty integration of AI into Bing has not lifted the unpopular search engine. A clumsy launch of Apple Intelligence could yet squander Apple’s strong position through similar flaws, and, in a severe case, could arrest the entire industry’s momentum.

Privacy may also be an obstacle. Consumers are accustomed to sacrificing privacy for capability. But a tool that could operate independently with a user’s data may be a step too far. Apple has suggested that it will silo all user data. But the company has no history of offering LLMs to consumers, and regulation to date has not been sufficiently focused on data transparency.

It may well be that a competitor’s AI proves its viability ahead of Apple. Google’s computational lead and suite of office applications could push the Android ahead of the iPhone. Or perhaps Amazon’s popular Alexa will overpower Siri — making this announcement just one of many in the AI race.

Whether Apple’s assistant dominates the market or not, the announcement of its AI strategy has made the company the world’s most valuable once more. Maybe this time the value attached to AI will live up to the hype.

Letter in response to this editorial comment:

Fears that AI will supplant humans are misplaced / From Graham Bell, Emeritus Professor, McGill University, QC, Canada