Apple's 2024 WWDC Shows Off How Evolutionary GenAI Is - Less Revolutionary

Apple may not have its own large language model (LLM), but it will be just fine. Apple has long tapped into artificial intelligence to drive improvements in experiences on all of our devices. It will do so with generative artificial intelligence (GenAI) and other emerging technologies that come along. Consumers appreciate these cumulative improvements even if their “wow energy” goes to the hardware with brighter screens and immersive experiences on a Vision Pro rather than the convenience of more invisible experiences – the use of AI to help me get things done more easily.

Apple has a position of strength. It has a lot of advantages when it comes to consumer digital experiences – including those that may be powered by- or enhanced with GenAI. They have 1) an installed base of more than 2 billion devices (e.g., hardware, operating system, apps) globally owned by consumers who are more affluent than the average citizen 2) vertical integration in their products that provides phenomenal seamlessness, and 3) a deep understanding of consumers on every dimension – something too many big tech companies lack. 

Here’s where Apple is really smart about LLMs and GenAI:

· Much of the value in LLMs is in the applications. Consumers will not have the ability to differentiate among LLMs. In the near term, brands will use GenAI to evolve existing apps (e.g., Email, word processing, search) and services in part because they can, but also because consumers expect software-based products to continue to add features and create better experiences. Soon we’ll see new apps and services for which GenAI is essential rather than a nice-to-have.

· GenAI is just one element in an experience – even if a very powerful one. Apple is tucking genAI features into its existing portfolio of apps and giving developers to the tools to do the same. It isn’t rolling out “GenAI Apps.” Even in applications that brand themselves with GenAI are often using more flavor AI than generative. It is an open question if consumers even care how spellcheck works or photo memories are created.

· Controlling access to consumers is essential. Apple’s ownership of its operating system (again on 2B devices globally) gives it control of who else has access to those consumers. It exercises this control in the spirit of delivering the best experiences while ensuring its consumers’ privacy and security. It also has a payments relationship with these consumers. About one year ago, Apple announced that it had paid out $1.1T to developers in 2022 alone.

Over the nearly 25 years that I worked as an analyst, I tracked 1) the use of personal context to deliver relevant experiences and 2) the large tech companies’ portfolios of consumer data, and 3) the arms race to become the consumers’ virtual assistant or agent. Big tech companies have harbored visions of playing that role for consumers for more than a decade – if not two. Apple showed that it understands what matters to consumers and how it will play this role. Here’s what I know about consumers along with the advantages Apple has: 

· Personal context matters and Apple may have more than any other entity. Personal context has a number of components. Two of the most important are: 1) a consumer’s time – what media I consume and 2) personal data (e.g., health, financial, location, etc.) When the first iPhone shipped more than 15 years ago, it shipped with 16 pre-installed apps. Now it ships with 30 pre-installed apps. Apple loyalists share data including: fitness, health, spending, location, friends/family, photos, media consumption, and more. My phone (or Apple) has the potential to know more about me than anyone – or even me if it can use its massive compute power to generate insights from all this context.

· Privacy matters – as does limiting the data available to third-parties for learning. Small LLMs on smartphones (expect Apple to announce this fall), private clouds, and data privacy policies should go a long way towards building consumers’ trust in using GenAI features in apps. Big tech companies have made billions or even trillions monetizing consumer data. Without arguing who has benefited the most – big tech or consumers -, consumers have told us that they want more control over how their data is used. They are also reading the press swarming the lawsuits from media outlets and backlash from artists whose work was used to train today’s LLMs. Even those consumers who shared their DNA may not have realized the business value they were creating for a third party.

· Consumers don’t buy technology – they want utility, entertainment, and more. Apple isn’t selling access to technology or a LLM. Nor did it use the term “productivity” that I remember during the keynote on June 10th. Apple’s primary story still focused on convenience, entertainment, and even some education (e.g., Math) and how GenAI can make that better. It will let consumers attach their subscriptions to OpenAI or other LLMs (eventually), but Apple isn’t tacking adding those charges to its consumers’ bills – yet.

· Consumers will continue to use many interfaces – not just conversational ones. More natural language interfaces are an “and” not an “or” to other options. Apple (in nearly its own words) launched the next era of Siri powered by GenAI. While one can use Siri with far more natural language for far more use cases (yep, the power of GenAI), consumers will still want screens and much more for entertainment, education, comparison shopping, photos, and more. Please keep in mind that consumers also have to be comfortable with conversational interfaces for more than communication. And communication isn’t always the most efficient way to get stuff done. Every heard a parent or manager use the phrase after talking to a child or employee, “oh, just forget it. I’ll do it myself.” I’ve heard some people describe the next generation of the Internet as just a microphone or a prompt box on a white screen. Neither is true – even though I do believe that websites will be mostly deserted in another 25 years (another conversation).

· Variable costs are not a risk consumers take on. Given consumers’ history of seeking flat rates independent of volume (e.g., SMS, Internet), I’d be surprised if consumers pay variable costs. LLMs (and now Apple) are testing consumers’ willingness to pay a premium for lower latency and higher quality results. So far, these are flat monthly fees just like they are for cloud storage, mobile phones, or streaming services. Many LLMs still or also have models with fees based on the capabilities of the model as well as the number of input and output tokens. Shorter prompts or outputs don’t necessarily deliver better results. Enterprises can manage this efficiency, but consumers are less likely to do so (even if Apple makes it easy by abstracting away the complexity of prompts to improve results).

Aside from Apple Intelligence, Apple announced dozens of improvements to experiences in the first hour plus. Most of these fell within their north star of giving consumers more personalized, relevant, secure, private, fun, intuitive, and convenient experiences that enrich their daily lives.