.png)
In a recent revelation, Apple has addressed rumors and clarified that its AI features, collectively known as Apple Intelligence, are not powered by the company's OpenELM AI model. This statement came in response to a report by 9To5Mac, in which the Cupertino-based tech giant emphasized that “OpenELM doesn’t power any of its AI or machine learning features – including Apple Intelligence.”
This clarification follows a Wired report that suggested major tech companies, including Apple, Nvidia, and Amazon-backed Anthropic, had used material from thousands of YouTube videos, including subtitles, to train their AI models. The report claimed that Apple utilized the plain text of video subtitles along with their translations into various languages for training its OpenELM model.
Understanding OpenELM
Apple has stated that OpenELM was created to contribute to the research community and to advance the development of open-source large language models (LLM). According to Apple, OpenELM is solely a research initiative and not used for powering AI features in Apple’s products and devices. This project aligns with Apple’s broader vision of enhancing AI research and promoting open-source collaboration, rather than serving commercial applications directly related to Apple Intelligence.
Google's Stance on YouTube Data Usage
The issue of using YouTube content for AI training is a contentious one, especially since Google explicitly prohibits the use of videos posted on YouTube for applications that are independent of the video platform. This policy aims to protect the integrity and privacy of content creators on YouTube, ensuring that their work is not exploited for purposes outside of the platform’s ecosystem.
Apple's Commitment to Data Privacy
In a research paper published on June 10, Apple reiterated its commitment to user privacy, stating that it does not use private personal data or user interactions for training its AI models. Instead, Apple uses publicly available data from the web, collected through its web-crawler AppleBot. This approach allows Apple to respect user privacy while still advancing its AI capabilities. Web publishers have the option to opt out if they do not wish to allow Apple to use their content for AI training.
The Scope of OpenELM
In April, Apple made its OpenELM AI model available on the Hugging Face model library. OpenELM, short for "Open-source Efficient Language Models," comprises a series of four small language models designed to run on devices such as phones and PCs. These models range in complexity, featuring 270 million parameters, 450 million parameters, 1.1 billion parameters, and the largest with 3 billion parameters. Parameters in AI models refer to the number of variables an AI system can use from its training data to make decisions.
For context, Microsoft's Phi-3 model can handle up to 3.8 billion parameters, while Google's open model Gemma, launched earlier this year, supports up to 2 billion parameters. This comparison highlights the competitive landscape of AI development and the significant resources companies are investing in advancing their AI technologies.
Implications and Industry Impact
The clarification from Apple is crucial in maintaining trust and transparency with its users and the broader tech community. By openly discussing the purpose and scope of OpenELM, Apple aims to dispel any misconceptions about its AI training methods and reinforce its commitment to ethical AI practices.
Furthermore, this situation underscores the ongoing debate about data usage and privacy in AI development. As AI technologies continue to evolve, companies must navigate the fine line between innovation and user privacy. Transparent communication and strict adherence to data privacy principles will be essential in gaining and retaining public trust.
Conclusion
Apple's recent statements provide valuable insights into the company’s approach to AI development and data privacy. By clearly distinguishing the roles of Apple Intelligence and OpenELM, Apple seeks to reassure users that their personal data remains protected and that the company’s AI advancements are grounded in ethical practices. As the AI landscape continues to grow and evolve, such transparency will be key in fostering trust and promoting responsible innovation.
コメント