Apple GPT: Everything You Need to Know
With the skyrocketing popularity of generative AI tools like ChatGPT, rumors have been circulating about Apple’s own AI product, potentially named “Apple GPT.” In this comprehensive guide, we dive into Apple’s experiments with AI and provide regular updates on the latest rumors. Keep reading to learn more.
Internal Testing and Development
Leading the development of large language models within Apple is AI chief John Giannandrea, who reports directly to CEO Tim Cook. Giannandrea established a team dedicated to conversational AI four years ago, and their work has been intensifying. Over the past few months, Apple has been testing a potential rival to ChatGPT under the internal name “Apple GPT.” This demonstrates Apple’s commitment to AI, as they have even designed their own framework called “Ajax” to support large language models. While the internal chatbot is referred to as “Apple GPT” by engineers, it should be noted that this name will not be used for a consumer product. Access to this chatbot is restricted to Apple employees with special permission, and it is strictly used for product prototyping and answering questions related to training data.
Ajax, equipped with over 200 billion parameters, is expected to outperform the previous ChatGPT 3.5. However, as of September 2023, newer models from OpenAI are reputedly more advanced than Ajax. Currently, Apple does not have a concrete strategy for creating a generative AI product targeted at consumers. Nonetheless, Apple is exploring various avenues, such as enhancing Siri, developing software to generate videos and images, and multimodal AI technology that incorporates images, videos, and text. The company is investing millions of dollars daily into conversational AI research, largely due to the resource-intensive nature of training language models.
AI for Siri
Apple’s work on generative AI has the potential to be integrated into its voice assistant, Siri, in the future. In February 2023, Apple organized an AI summit to educate employees about their progress with large language models. Due to the fundamental structure of Siri, implementing new capabilities can be time-consuming, as it requires updating the existing database. John Burkey, a former Siri engineer, highlighted the challenge of adding new phrases, which necessitates rebuilding the entire Siri database, a process that can take up to six weeks. Similarly, more complex features like search tools may take over a year to implement. Furthermore, Apple must address privacy concerns before incorporating generative AI into Siri. While competitors like Alexa and Google Assistant have prioritized functionality over privacy, Apple has taken a more cautious approach, resulting in Siri lagging behind its counterparts.
According to The Information, Apple is working on a feature that would enable Siri to automate multi-step tasks, similar to the existing Shortcuts app.
Third-Party AI Bans at Apple
Genuine concerns about data confidentiality have led Apple to prohibit the use of AI tools like ChatGPT and GitHub Copilot among its employees. Apple fears that these tools could potentially leak sensitive company information. This cautious stance is also adopted by other entities, including government agencies, healthcare institutions, and banks, who have barred their employees from using such generative AI tools. Even major tech companies such as Verizon, Google, and Samsung have implemented similar bans.
AI Competition
Nearly every major tech company is actively involved in developing AI products. Besides ChatGPT from OpenAI, Google, Microsoft, and Amazon have all incorporated generative AI into their offerings. Google has created Bard, a chat-based AI tool, and integrated generative AI into its search products and apps. Microsoft has partnered with OpenAI, integrating ChatGPT with its Bing search engine. Amazon aims to enhance its voice assistant, Alexa, using generative AI. Moreover, Meta (formerly Facebook) has integrated generative AI into apps like WhatsApp, Messenger, and Instagram for various functionalities.
Apple’s Current Use of AI and Machine Learning
Apple is already leveraging artificial intelligence and machine learning in numerous ways across its product lineup. Here are just a few examples:
- Photos – Machine learning improves photo quality on iPhones through features like Deep Fusion and Portrait Mode adjustments.
- Spotlight Search – AI powers search functionality throughout the iOS operating system.
- Visual Lookup – Machine learning enables content recognition in photos.
- Crash Detection and Fall Detection – Machine learning is used to analyze sensor data and detect potential accidents.
- ECG – AI examines heart rhythm data to identify signs of atrial fibrillation on Apple Watches.
- Autocorrect – Machine learning algorithms drive Apple’s autocorrect system and word suggestions.
- Siri Suggestions – Machine learning algorithms make intelligent recommendations across various apps on iPhones.
Apple’s devices are equipped with dedicated Neural Engines that enhance their machine learning capabilities.
Tim Cook on AI
Apple CEO Tim Cook has expressed keen interest in AI, acknowledging its potential while emphasizing the need for careful consideration. Cook believes that AI and machine learning are foundational technologies for all of Apple’s products and plays a critical role within the company. He confirmed that Apple has been experimenting with generative AI for several years.
Potential Launch Date
Both The Information and analyst Jeff Pu predict that Apple may introduce a generative AI feature for the iPhone and iPad in late 2024, coinciding with the release of iOS 18. Pu indicates that Apple plans to deploy hundreds of AI servers in 2023, with further expansion in 2024. The company intends to offer a combination of cloud-based AI and on-device processing.
However, Apple analyst Ming-Chi Kuo believes that Apple lags significantly behind its competitors in generative AI development, suggesting that a 2024 launch may not be feasible.
Guide Feedback
Do you have questions about Apple GPT or want to provide feedback on this guide? Feel free to reach out to us via email here.
