Apple shared some interesting insights about its Large Language Model (LLM): ReaLM. With ReaLM, Apple aims to transform our everyday interactions with our devices, offering a smarter, more intuitive user experience that seamlessly integrates with our daily lives. The Cupertino firm keeps investing in AI advancements, as evidenced by this research publication.
Apple’s ReaLM competes with other LLMs
Apple’s AI experts outline a system that enables Siri to go beyond mere image recognition capabilities. They also claim this system not only competes with but surpasses GPT-4, based on their internal benchmarks:
| Model | Conv | Synth | Screen | Unseen |
|---|---|---|---|---|
| MARRS | 92.1 | 99.4 | 83.5 | 84.5 |
| GPT-3.5 | 84.1 | 34.2 | 74.1 | 67.5 |
| GPT-4 | 97.0 | 58.7 | 90.1 | 98.4 |
| ReALM-80M | 96.7 | 99.5 | 88.9 | 99.3 |
| ReALM-250M | 97.8 | 99.8 | 90.6 | 97.2 |
| ReALM-1B | 97.9 | 99.7 | 91.4 | 94.8 |
| ReALM-3B | 97.9 | 99.8 | 93.0 | 97.8 |
This development suggests a future where Siri becomes more intelligent and efficient. Apple appears optimistic about its execution of this project, showcasing its speed when benchmarked against OpenAI’s ChatGPT 3.5 and 4.0 models:
“As another baseline, we run the GPT-3.5 (Brown et al., 2020; Ouyang et al., 2022) and GPT-4 (Achiam et al., 2023) variants of ChatGPT, as available on January 24, 2024, with in-context learning. As in our setup, we aim to get both variants to predict a list of entities from a set that is available. In the case of GPT-3.5, which only accepts text, our input consists of the prompt alone; however, in the case of GPT-4, which also has the ability to contextualize on images, we provide the system with a screenshot for the task of on-screen reference resolution, which we find helps substantially improve performance.”
Super-brain Siri
In its research, Apple introduces ReaLM as a significant enhancement for Siri, essentially equipping it with a “super brain.” This upgrade significantly improves Siri’s ability to comprehend user requests by analyzing on-screen content, such as text messages and images, to anticipate users’ needs more accurately. ReaLM also keeps track of ongoing conversations with Siri, allowing it to recall previous mentions for better assistance.
ReaLM essentially acts as an omnipresent observer, capable of detecting background activities like alarms or music and tailoring its interactions accordingly.
The research includes a sample dialogue showcasing its capabilities.
| Speaker | Dialogue |
|---|---|
| User | Show me pharmacies near me. |
| Agent | Here is a list I found. |
| Agent | … (list presented) |
| User | (eg 1) Call the one on Rainbow Rd. |
| User | (eg 2) Call the bottom one. |
| User | (eg 3) Call this number (present onscreen). |
When compared with other intelligent systems, ReaLM has demonstrated superior problem-solving skills, making it a standout feature that operates directly on iPhones and other Apple devices. This approach ensures that users’ data remains secure and that the system operates with exceptional speed, enhancing Siri’s intelligence and understanding of users’ actions and speech.
While there’s no confirmed release date for Apple’s ReaLM, it’s anticipated that aspects of this technology might be integrated into upcoming versions of Apple’s operating systems, such as iOS or macOS. Future events like the Worldwide Developers Conference (WWDC) in June may provide more insights into its implementation.
Featured image credit: Stephen L/Unsplash
