Why is Apple cautious about discussing AI?
After years of deep cultivation in AI, Apple is not concerned about how cool AI itself is, but rather focuses on the user experience brought by AI.
Nowadays, AI has become the center of attention in the tech industry, with almost all tech companies diving into it. However, Apple has always been cautious about AI and even didn't mention it at all in the recent WWDC.
What are the reasons behind Apple's caution towards AI?
In a recent article by InfoWorld, Matt Asay analyzed this issue and pointed out that "Apple's focus is on user experience, and for Apple, what's important is not AI, but the improvement of user experience brought by AI."
Matt Asay is currently responsible for marketing at the open-source database technology company MongoDB. Prior to that, he was in charge of Amazon Web Services and Adobe's developer ecosystem.
Here is the original article:
You are not alone in being overwhelmed by the AI wave, everyone is caught up in it. Developers can't stop talking about how GAI is changing the way they program, and CEOs can't stop talking about it either. Recently, AI has been deeply intertwined with financial conference calls, and we have even started using the output of GAI applications to train the large models (LLMs) behind these applications. (This won't end well).
However, in all the endless hype and anticipation surrounding AI, there is one company that recently spoke for more than two hours at a global keynote speech, and even though AI is ubiquitous in its products, it never mentioned AI once. That company is Apple, and its cautious approach to speech has taught us all a lesson on how to use AI correctly.
Easier said than done
Although discussions about artificial intelligence are increasing, as Gartner Vice President Analyst Mark Raskino pointed out, the hype surrounding AI is nothing new. Discussions about AI began in the early 1980s and have not slowed down since. What's different now is that AI has become very common both inside and outside the tech industry. For some, ChatGPT has become a "viral hype."
This speed has had some consequences, such as the fact that every LLM requires data, and rich sources of data such as the Internet Archive, Stack Overflow, Reddit, and others have seen a surge in traffic, leading to the collapse of the Internet Archive and the blocking of Reddit. At the same time, some are opposing so-called copyright infringement by using it to train applications such as GitHub Copilot, which is all a bit chaotic.
In fact, as James Penny, Chief Investment Officer of TAM Asset Management, suggests, companies that even mention the word AI in their profits will see their stock prices rise, much like the early days of the Internet. Although it seems a bit foolish considering that things like GAI are still very primitive, there is evidence that AI has driven the prosperity of the stock market, but has not actually played a significant role in driving corporate profits. Indeed, there is a bit of a "dot-com bubble era" feel to it.
At the same time, there is a company that is constantly investing heavily in AI, but has not made a big deal about it. That company is Apple, which uses AI more responsibly and effectively than most companies.
Behind the Scenes Efforts
AI is not unfamiliar to Apple, which has already made AI a part of its products through Siri and other less obvious ways. It is no surprise that Apple has long been recruiting AI talent, and this recruitment has become more prominent recently. The company has a landing page specifically for artificial intelligence, with the title "Machine Learning and AI: Work is Innovation, Experience is Magic."
On this page, Apple reveals its secrets on how it applies AI, "People working in machine learning and AI here are creating magical experiences for every Apple product, enabling millions to do things they never imagined possible." In other words, their focus is on how customers experience AI, not AI itself. This has always been Apple's approach: to make technology a part of the customer experience, but not the focus of that experience. The purpose of this technology is essentially invisible, and if you notice it, Apple has failed.
On the stage of its annual Worldwide Developers Conference, Apple tends to refer to AI as magic. The word was used 13 times. Speaking of Apple's latest Vision Pro, Apple executive Alan Dye praised it, "It's amazing, it feels like magic." He doesn't need to delve into the details of artificial intelligence and other technologies, which provide support for this magic. The focus is on the experience, not the technology.
This is a good lesson for every company.
First of all, although GAI is the current technology, it is not always the right approach. Mathew Lodge of Diffblue recently suggested that in some use cases, reinforcement learning is better than GAI. Even before GenAI became a hot topic, it was believed that regression analysis or other methods should be the first stop for companies before they embarked on machine learning.
Recently, I had a conversation with an industry friend who emphasized, "You can do a lot with LLM, but if your output is structured data rather than unstructured data, it may be a very inefficient way to do it." This is an intriguing point, as some of the GenAI services that cloud computing providers are rolling out have already been done with dedicated models, which have proven to be more efficient than GenAI. As he explained, developers are fascinated by GAI because it is essentially probabilistic. It is not about finding the one true answer, but about finding a reasonable answer in the training data. This may be a good thing, but it's like searching without an index. Its scale is not large.
This does not mean that GAI is bad. It is just not the best or even good in a series of use cases (even if it is a great approach, it still requires a lot of resources). For some use cases, the old-fashioned reasoning method works best. Reasoning is a way to train AI to see patterns in data and then compare incoming new data with these patterns. Similarly, GAI creates something that looks like data in LLM, resulting in newly created data that is reasonable but not necessarily correct. Both are interesting but not necessarily correct tools.
Second, no matter which AI approach a company chooses (the reality is that most companies will want to accept a range of approaches because they will have a range of use cases), AI should never be the focus. AI is a means, not an end. As Apple has shown, it is very possible to promote the vision of AI without making it the focus of advertising. No one really cares how cool AI is, they care about the experience it brings.
So, Apple sells the experience of AI, not AI itself.