Apple, a company with a capitalization of 2,700 billion, has long started a real hiring spree to expand its team engaged in the fields ofartificial intelligence he was born in machine learning.
According to an investigation by Financial TimesApple would be engaged in an action of recruiting that involved at least 36 Google specialists in the field of AI. And although a large part of the team works in the Californian and Seattle (USA) offices, the giant would have created a secret laboratory in Zurich, Switzerland. It would be the result of the acquisitions made by Apple in Switzerland: known as Vision Labthe new location of the company led by Tim Cook, was born after the purchase of companies such as FaceShift (virtual reality) e Fashwell (image recognition).
Apple committed to developing even more advanced generative models that work locally
Sources close to the US multinational suggest that Apple is busy developing innovative generative models capable of running on devices with an overall modest hardware configuration, rather than on servers located within data centers.
On the other hand, the news of the launch of OpenELM, an open generative model that works on iPhone and iPad, dates back to just a few days ago. While preferring not to reveal any technical information, the CEO himself Tim Cook stated that Apple “is conducting research on a wide range of artificial intelligence technologies” and is investing and innovating “responsibly” around the new solutions under development.
One of Apple’s primary objectives would be to evolve theSiri digital assistant migrating it towards a neural architecture for speech recognition. Chuck Wootersexpert in conversational artificial intelligence and Large Language Models (LLM), has been working on improving Siri for almost two years, starting in December 2013. Already so many years ago, Apple’s top management were big supporters of neural networks, well before LLMs began to establish themselves and show their incredible potential. Consider that the document “Attention Is All You Need” by Google is from 2017 (Transformers concept) while the chatbot ChatGPT of OpenAI (with the underlying GPT model, Generative pre-trained transformer) is November 2022.
How Apple’s artificial intelligence team is made up
Apple’s AI team is now made up of former key Google figures, among which it stands out John Giannandreawho was previously responsible for Google Brain, the artificial intelligence laboratory that later became part of DeepMind.
Both Bengiosenior director of AI and ML research, was previously one of the best scientist Google. Ruoming Pang, who leads the team of “Foundation Models” of Apple, working on LLMs, has previously led the research regarding the AI speech recognition by Google.
Apple had also hired Ian Goodfellow, another pioneer in the field deep learning. However, he returned to Google in 2022, criticizing the policy followed by Apple.
Il Financial Times it also mentions the names of other former employees of the Mountain View company that Apple wanted in its ranks. Among them are the authors of a major research paper published in March 2024: Apple revealed that it has developed a family of artificial intelligence models known as “MM1” that use textual and visual inputs to generate responses.
“The next big thing will be ‘AI smartphones’ – and these will require a lot more DRAM“said Sumit Sadana, executive vice president and chief business officer of Micron Technology, one of Apple’s chip suppliers. “To run LLM, the minimum required is 12 GB of RAM“.
The next iPhones could integrate much more than a voice-activated intelligent personal assistant. The next evolutions of Siri for example, they could interact with all the apps on the phone via voice control. Already during the next WWDC, one or two artificial intelligence features are expected that can become revolutionary elements for the average consumer.
Opening image credit: iStock.com – BlackJack3D