Apple AI Researchers Share Possibility of Outperforming GPT-4

Apple AI researchers reported a system that could help Siri perform more tasks than recognizing the details of an image which has shown better benchmarks than ChatGPT 4.0.

In the paper Reference Resolution As Language Modeling (ReALM), Apple shared an LLM-enhanced voice assistant that could have boosted features in the future.

Apple Tests System on GPT Models, Reveals Superb Performance

According to the research, the researchers have been doing demonstrations of its existing system with similar functionality across different types of references. Apple shared that they have tried benchmarking against the GOT-3.5 and GPT-4.

"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," Apple explained.

Moreover, the company is looking forward to developing a system that could exist on the device without compromising on performance.

Apple Teases on Smarter Siri in the Future

Apple has just started to embrace the AI hype this year by hinting at possible integration for its next major upgrades. For instance, Siri, Apple's voice assistant, could be the first one to receive AI features.

"... with our smallest model achieving performance comparable to that of GPT-4, and our larger models substantially outperforming it," the company wrote. Hence, the researchers concluded that ReALM is an ideal choice for a practical reference resolution system.

Apple has been reportedly getting more proactive on AI integration for its upcoming models. With the development of ReALM, it does not seem farfetched anymore.

Apple enthusiasts are now hoping to start seeing the first batch of the development on iOS 18 at the Worldwide Developers Conference on June 10.