Last year, Apple Intelligence was introduced. It included a next generation Siri with features like a new screen animation, recognition of more natural speech, conversational context, enhanced product knowledge, on-screen awareness, inapp actions, and personal context. But now we know the only thing that actually functioned in that demo was the colorful Siri animation around the edge of the screen. Everything else was faked. In fact, members of the Siri team were shocked by the demonstration since they'd never seen those features running on a test device. That means Apple lied about what Siri could do, and they crossed their fingers that by fall 2024, they'd have it figured out. But that didn't happen. Instead, they've delayed the upgraded Siri until late this year or even next year in 2026. Now, you'd think Apple would have learned their lesson after Air Power, a charging mat revealed in 2017, scheduled to be released in early 2018 until it was never mentioned again until 2019 when Apple officially cancelled the product. Lesson being, don't introduce something that isn't ready. But Apple did something even more egregious with Apple Intelligence. They advertised it alongside their new products with iPhone 16, 16 Pro, 16e, MacBook Air, iPad Air, MacBook Pro, Mac Studio. They plastered it on everything they possibly could despite knowing full well they were advertising a feature that literally didn't exist and likely wouldn't for years now. If that sounds illegal, that's because it is. And Apple's currently facing a lawsuit for it from the US District Court in San Jose, claiming that the company's promotion of now delayed Apple Intelligence features constituted false advertising. And it's worth pointing out that this is uncharacteristic for Apple historically. The original iPhone was famously demonstrated live on stage with no trickery. And the company rarely previewed new products or features that weren't ready to be shipped within a few months. So why did Apple change their approach with Apple Intelligence and Siri? Well, it's because they face so much pressure to catch up in the AI race. So they made a big promise that they hoped would be possible to deliver, but it's turning out to be more difficult than expected. In fact, Apple wasn't even sure what type of technology they'd need to make Siri work. And this might be because they weren't the company who actually created Siri. It began as an app that Apple purchased, but the creators of the app said it wasn't really designed to do everything Apple wanted. Because Siri wasn't really an AI. It was more of a voice recognition technology. That's why one of its original creators went on to form a new company called Viv that was closer to the large language models we're used to today, like for example, Chat GPT, which makes Siri feel archaic in comparison. So, why didn't Apple just buy Viv or OpenAI or some large language company that has more modern AI technology? Well, it's likely because Apple genuinely believed in their development plan for Siri. It just didn't work out. It all began in 2018 with John G and Andrea. He was in charge of artificial intelligence at Google and at one point was considered the most sought-after tech leader in Silicon Valley. So, bringing him on board at Apple appeared to be a big win for the company. He believed Siri could be fixed by using the same strategy as Google to create their AI voice assistant, which was quality training data and better web scraping to answer general knowledge questions. But those were marginal improvements, and Siri needed a complete overhaul, which included its team. Many at Apple told Geandria he should replace and reorganize Siri's leadership, but he didn't. One of those leaders was Robbie Walker, who became Geian Andrea's right-hand man instead of being demoted or fired. Walker had led the Siri team since 2013 and developed a poor reputation inside Apple since he never took any big risks or laid out any ambitious plans for Siri. In fact, his crowning achievements were reducing the time it took Siri to answer a question and removing hey from the hey Siri voice command. Meanwhile, Google's AI assistant can remember what you said in previous conversations, handle multiple commands at once, and answer complicated questions without redirecting users to the internet. Walker also stopped engineers from utilizing large language models to give Siri a deeper understanding of more topics. Instead, he felt focusing on the next incremental Siri update was a more important use of resources. No one's sure why Jane and Andrea kept Siri's previous leadership team intact, but it resulted in virtually no improvement to the product's poor performance. Now, as you can imagine, this upset other Apple executives who were actually good at their job, like Craig Federigi and Mike Rockwell. Federigi was in charge of the software team while Rockwell led development of Apple's Vision Pro headset. They're known among engineers as having demanding type A personalities while Gianandrea was relaxed, quiet, and non-confrontational. During meetings, he preferred to sit and listen, while Federigi asked most the questions. He also had a very hands-off approach to managing his employees. While Federigi often knew more details about software projects than the engineers working on them. Over time, Fedri's software team and Geianandrea's Siri team began to clash and resent each other. Not only because of their opposite leadership styles, but because AI engineers were receiving faster promotions and higher pay than the software engineers despite having less work output. In fact, Federi became so frustrated with the Siri team's slow adoption of new software features that he ended up creating an artificial intelligence team of his own referred to as intelligence systems. It was made up of hundreds of machine learning engineers who developed language models that allowed users to control apps with voice commands. The Siri team's slow productivity frustrated both Federigi and Rockwell, who wanted Vision Pro users to browse the internet and resize windows using their voice. But Gene Andrea's team made very little progress in supporting that functionality, forcing the concept to be dropped from the final product. Now, when chat GPT came out in 2022, Federigi's team took the technology seriously and immediately began researching large language models, figuring out ways it could improve the iPhone experience. But Gene Andrea's team didn't have the same sense of urgency. Initially, the intelligent systems team created demos for Federigi using Open AI's LLMs, showing how AI could understand what was on a user's screen, recognize more natural speech, and handle multiple commands at once. But by 2023, Apple told their engineers they had to stop using outside models to create new features and rely solely on their own in-house models. The problem was Apple's technology wasn't nearly as good as open AIs. At the same time, Apple faced pressure from shareholders to introduce their own AI features since Siri had already failed to compete with voice assistants from Amazon and Google. While Open AI had raised the bar even higher with Chad GPT, Apple also needed to give users a reason to upgrade their devices since iPhone, iPad, and MacBook hardware had become stagnant in recent years. So, introducing a bunch of new AI features that are only available on the latest products would reassure shareholders and likely boost product sales. That's why on June 10th, 2024, Apple decided to lie. They demoed a next generation Siri powered by Apple's new AI technology that literally didn't exist. And this lie paid off in the short term. Apple's stock price jumped 10% 2 days following the announcement. And they could advertise their AI features alongside upcoming product releases. But in the long run, customers trust in Apple eroded. There was confusion in fall 2024 when the Siri animation changed but worked just as bad as before. Then in spring 2025, when Apple originally promised to deliver their AI features, but were delayed again. This time, the release window became even more vague, saying the new Siri would arrive quote in the coming year, which could mean tomorrow or March 2026. Now, Apple has taken measures to get the feature introduced by the end of this year, like replacing John G and Andrea with Mike Rockwell as head of the Siri team just last month. While Federi began allowing Apple's machine learning engineers to use open- source large language models from other companies, which is a big shift in strategy and may mean Apple's willing to compromise on user privacy in order to speed up series development. But no matter when the new series is released, it better function exactly as Apple demonstrated, or else they risk even more damage to their reputation while falling further behind in the AI race. This is Greg with Apple Explained. Thanks for watching till the end, and I'll see you in the next video.