According to the latest report, Apple seems to have found a way to surpass ChatGPT, opening up a new era in the artificial intelligence race. The report states that the company is developing a Large Language Model (LLM) called ReALM, which stands for Reference Resolution As Language Modeling.
The main goal of the ReALM model is to enhance Siri's contextual understanding abilities, as announced in a research report by Apple. This information was first shared by Windows Central, revealing the advancements that Apple has made in the AI generation development race.
ReALM is not just a significant advancement for Siri but also a strong statement from Apple in the AI field. This model has four different variants, and even the smallest variant is said to operate at a level comparable to OpenAI's ChatGPT.
Compared to ChatGPT, which can handle image and PDF files, ReALM is said to be capable of processing on-screen content and providing responses based on that content. This could create a significant competitive advantage for Apple.
Apple introduced ReALM with four different sizes: ReALM-80M, ReALM-250M, ReALM-1B, and ReALM-3B, equivalent to 80 million, 250 million, 1 billion, and 3 billion parameters, respectively.
Apple's researchers evaluated ReALM against OpenAI's GPT-3.5 and GPT-4, and the results, not surprisingly for a company backed by Microsoft, showed that even the smallest ReALM-80M model from Apple achieved performance on par with GPT-4, while larger models surpassed it by a significant margin.
ReALM has demonstrated significant improvements compared to similar function systems, with the smallest model outperforming by up to 5% in processing on-screen information.
The exact timing of Apple's LLM model launch into the mainstream market is still unclear. However, it is possible that Apple will discuss ReALM at the upcoming WWDC event in June.