06/17 2025
512
Author: Lin Yi
Editor: Key Point Jun
Following Apple's WWDC2025, software chief Craig Federighi and marketing chief Greg Joswiak faced the media, addressing the company's most scrutinized question: the status of Apple's AI initiatives.
Last year, Apple unveiled plans for a smarter Siri, yet the promised upgrade failed to materialize, leading to speculations of a marketing ploy.
How did Apple's leaders respond? What caused Siri's delay? This interview unveils crucial insights:
1. Behind the Delay: A Quality Quandary
Federighi admitted Siri's delay, explaining, "We had a two-phase plan. We had a usable version, but it didn't meet our quality standards." Joswiak added, "We refuse to release a flawed product, as that would disappoint our users."
This reveals Apple's stance on AI: prioritizing quality over speed. Even amidst market pressure, they won't rush an unfinished product.
2. Apple's AI Vision: Seamless Integration
Joswiak emphasized that Apple aims to embed AI deeply into the OS, making it an invisible yet powerful tool. Federighi compared it to the internet's early days: "No one said, 'I'm using the internet,' but they used it for searches and shopping. AI should be the same—an underlying capability."
3. Quality Over Speed
Federighi stressed, "We don't need to be first; we need to be the best." Apple prioritizes refining models before release, unlike competitors like OpenAI. Joswiak admitted, "AI isn't easy, especially when considering device-side running, privacy, and seamless integration."
4. Siri's Evolution
Siri, launched in 2011, once led the industry but has since fallen behind. Federighi said, "We aim for Siri to be indispensable, but we're not there yet. We won't give false hopes."
5. System and Design Updates
Apple also discussed iOS 26's liquid glass design and potential foldable iPhones. Joswiak dodged the latter question, saying, "Who knows? We haven't seen one."
6. Apple's AI Strategy
Amidst the AI rush, Apple chooses a slower, more deliberate path. Federighi reiterated, "Our goal is not to be first, but to build the right products." This approach may prove pivotal as companies seek to integrate AI into users' lives.
Below is the full transcript of the interview:
I. AI and Siri Delay
Host: Craig, Greg, thank you for speaking with us at this significant developer conference.
Federighi: Thank you for having us.
Host: Yesterday was pivotal. I'd like to discuss last year's WWDC announcement of a smarter Siri. Where is it now?
Federighi: We spent about 40 minutes on Apple Intelligence, with eight minutes focused on Siri. We delivered some Siri updates but not all planned features. We laid the groundwork for Apple Intelligence, including a semantic index and large language model on-device, driving many functions. We had a two-phase plan for Siri, but the first version didn't meet our quality standards. We aimed for reliability, which we couldn't achieve within our timeframe.
Host: So, there was a usable version, but it wasn't ready.
Federighi: Yes, it was real software with a large language model and semantic search capabilities. We anticipated launching it later this year but realized it wouldn't meet our standards.
Joswiak: As Craig said, we thought we could launch by year-end. Our marketing reflected that. But we wouldn't disappoint customers with a flawed product. The error rate was unacceptable, so we made the best decision.
Host: How did the marketing happen?
Joswiak: We planned to launch by year-end, so we showcased Apple Intelligence features, including one with Bella Ramsey. Most of our marketing focused on Apple Intelligence's capabilities.
Federighi: Apple Intelligence encompasses 20 features. If you read last year's event coverage, you'd think it was all about Siri, which was only 4% of our announcements. We heavily promoted Apple Intelligence as a great foundation for the future, not just a chatbot.
Host: With your resources, why couldn't you deliver?
Federighi: It's new technology. Nobody's perfectly implemented device-based AI yet. We want to do it best. We have promising results, but they're not reliable enough for an Apple product. AI is a long-term wave of change, and we want to get it right. There's no need to rush just to be first.
Host: Siri is synonymous with Apple's AI. People expect it to excel.
Craig Federighi: Ultimately, we aspire to reach that pinnacle, but we're not quite there yet. It's our mission, and we're fully committed to it. Last year, we shared our direction, and people have high expectations for Apple's values. They want an experience that's seamlessly integrated into their lives, not just a chatbot on the side. It needs to be personal and private. We've started building and delivering some of these capabilities, including our excellent semantic photo search. Everything we do is directly integrated into how you use your device, and people are excited about this future. They're eagerly anticipating the next generation of Siri, and we're eager to deliver it—but we want to do it right.
Craig Federighi: In this case, we want to ensure everything is perfect before discussing specific dates. For obvious reasons.
Craig Federighi: Absolutely. I'd love to talk about these things, but it's crucial not to set the wrong expectations. We want to deliver something truly remarkable that you and all our customers will appreciate. You mentioned Apple Intelligence a lot.
Greg Joswiak: It's important to understand that our strategy is a bit different. Our philosophy with Apple Intelligence is to use generative AI as an enabling technology that powers various features in our operating systems. Sometimes, you're using a feature without realizing it's powered by AI. That's the goal—to integrate it seamlessly into your daily life to enhance your experience without you even thinking about it. There's no app called "Apple Intelligence," and it's not a chatbot. Some people get confused and ask, "Where's your chatbot?" We're not doing that. Instead, we offer access to a chatbot through ChatGPT because we believe it's the best. But our idea is to integrate AI throughout the operating system, making it a feature you use every day.
Craig Federighi: AI is a transformative technology, like the internet or mobile internet. When you look back at the internet, no one asked, "Why doesn't Apple have its own web search?" The internet is vast, and it's an opportunity for many companies. It enables users to do various things and is a huge driver for Apple. We've made the internet more accessible in many ways. But that doesn't mean every experience has to happen within Apple or ultimately on Siri. There are places where Siri will be a crucial, indispensable part of your daily life, but using other tools for other tasks isn't a problem.
Craig Federighi: We already are. Today, when you use writing tools to make content more concise, turn it into a bulleted list, create a table, or give temporary commands, like writing a haiku, those are all done using Apple models in the private cloud. We're increasingly using our models, including for some visual intelligence experiences. But ChatGPT offers excellent capabilities, so we've integrated it more to do powerful things. As you pointed out, it's now available for visual intelligence, not just for what you see with your camera but also for things on your screen.
We're also integrating other tools. We provide ChatGPT's image generation in Image Playground and various apps, both Apple and third-party. You can now access ChatGPT's image generation directly while using those apps. We also announced exciting coding tools in Xcode, which our developers loved yesterday. You can use ChatGPT and other models like Anthropic for writing code in Xcode, which is fantastic. We'll continue enhancing our own models but also want to ensure our customers have access to the best of everything.
Craig Federighi: We already are. In fact, we published a paper this year describing our latest on-device and private cloud computing models, both of which have undergone significant architectural improvements. Our private cloud computing model, PCC, is much larger and powerful, at GPT-4o level. But our goal isn't to create another chatbot.
Greg Joswiak: It's to integrate these capabilities in various ways across our operating systems and features. That powerful intelligence behind those features.
Craig Federighi: We do. It's an excellent material.
Craig Federighi: Glass has some useful properties for user interfaces, especially the adaptive glass we can make and incorporate into products. As displays get more advanced, larger, and with rounded corners, we want content to feel more open, taking up the whole screen. Glass allows us to embed control elements concentrically, making them almost invisible while still defining clear controls. It's also adaptive, transmitting background through translucency while providing contrast to see both the content behind and in front. It has great properties as part of a user interface and looks cool.
Craig Federighi: Hardware has evolved, including processing power, enabling the glass to transmit content like refracting it away from edges. It requires a lot of computation, but Apple silicon provides the powerful capabilities to enable this across our product line. We also have amazing displays, some with high-definition or HDR, allowing spectral highlights on the glass. Larger screens and the need for concentricity drove this development. Combining all these factors allowed us to take a big step forward.
We also saw the opportunity to bring a consistent experience across our product line. It took years to converge iOS and macOS with unified font types and design language. This year, we finally delivered on it.
Greg Joswiak: It's worth mentioning that this was inspired by years of work on Vision OS. What we learned and how much users loved it brought inspiration to consider this problem at a design level.
Greg Joswiak: I don't know. Who knows?
Greg Joswiak: Both are very popular and do great things.
Craig Federighi: They're both great. I love them both. The Mac is optimized for precision with a keyboard and trackpad, while the iPad is the ultimate touch device. As iPad screens get larger and more powerful, customers want to do more with it. We've found that some Mac expressions translate well to iPad, so we want to be consistent where it makes sense but also embrace iPad's unique aspects.
Greg Joswiak: The answer is no.
Craig Federighi: Exactly. Are there any Macs with touchscreens in those buildings? I don't see one.
Craig Federighi: It's a multimodal future. We're visual creatures who also have language and speech capabilities. You'll want to use all these capabilities simultaneously—speaking, touching, interacting. The ultimate future is one where you're watching, interacting, touching, and speaking, and the device can take your commands and show you what you want, just like we interact with the natural world.
Craig Federighi: I don't know what that is.
Greg Joswiak: Neither do I.
Craig Federighi: We have personal wearable devices. If you want something aware of your surroundings with audio, you're wearing one on your wrist. If you want something to capture and receive visual content, you probably have one in your pocket. Are there other form factors suitable for AI? Sure, but it's hard to beat something that's with you all the time and provides a nice screen to interact with.
Greg Joswiak: Obviously, we monitor this situation daily, but there's no news to share at the moment. We're all closely watching how things unfold. This week, our focus is on the exciting announcements across all our platforms, including Apple Intelligence.
Craig Federighi: Those software upgrades are free and not subject to tariffs. That's accurate, and you're hearing it from us first.
Host: But regarding marketing, if you had to raise prices, it must be challenging to promote a more expensive iPhone.
Greg Joswiak: It's too early to discuss that. There's really nothing to say today.
Host: So, you don't have a team working on potential pricing strategies?
Greg Joswiak: As you might imagine, we keep a close eye on it. It would be irresponsible not to. But there's nothing to announce right now. Okay.
Host: Together, you both have about 60 years of experience at this company. You've seen the company through its highs and lows. Where do you think the company stands now?
Craig Federighi: Many people think Apple is in a defensive position. You raise a good point, because when you go through different phases, you get accustomed to the ups and downs. I feel very good about where we are.
Greg Joswiak: Yeah, I don't want to sound naive, but I remember when Steve came back and said, "Our job is to make great products and tell people about them. If we do that, everything else will fall into place." And that's exactly what happened. We make great products, and we believe our products are excellent right now. They will continue to improve. We talk about them here, explain why they're great, and everything falls into place. Our business is strong. People love our iPhones. Macs are more popular than ever. iPad is growing and doing well. Everything we do is about making great products. Yesterday, you saw how we use technology, from design to AI and other functionalities, to elevate our products. Each of our flagship products ranks number one in customer satisfaction in their respective categories. We just want to keep improving them, and that's what we're doing. I still believe in that naive theory: if we make great products and tell people about them, everything else will fall into place.
Craig Federighi: Yesterday was truly exhilarating. We've been working hard on the new design and features. Presenting them to the audience and seeing their reaction, you can feel the energy, the passion, and the care people have for these platforms. You can feel the excitement about the improvements. And for us, we know what's coming next. We're all super excited about the great products we're building. It feels really good. That's all I'll say.
Host: Thanks for sitting down with me, guys.
Craig Federighi: Thank you.