Apple WWDC23 does not mention "artificial intelligence", preferring to use "machine learning"

IT House News on June 6, Apple, in addition to releasing the highly anticipated new products such as Mac Pro and Vision Pro, also demonstrated its latest advances in the field of machine learning during its WWDC 2023 keynote speech on Monday. progress. However, IT Home noticed that unlike competitors such as Microsoft and Google, which vigorously promote generative artificial intelligence, Apple did not mention the word "artificial intelligence" in its speech, but instead used "machine learning" and "machine learning" more often. Terms like "ML".

For example, in a demo of iOS 17, Senior Vice President of Software Engineering Craig Federighi described improvements to automatic error correction and speech recognition:

Automatic error correction is driven by on-device machine learning, and we have continued to improve these models over the years. The keyboard now leverages a transformer language model, the most advanced word prediction technology available, making automatic error correction more accurate than ever. And, thanks to the power of Apple Silicon chips, iPhone can run this model every time you press a key.

It is worth noting that Apple mentioned a term "transformer" in the field of artificial intelligence in its keynote speech. The company specifically talks about a "transformer language model," meaning its AI models use the transformer architecture, which is the underlying technology used by many recent generative AIs, such as the DALL-E image generator and the ChatGPT chatbot . The transformer model (a concept first proposed in 2017) is a neural network architecture for natural language processing (NLP) that uses a self-attention mechanism that allows it to prioritize different words or elements in a sequence. Its ability to process inputs in parallel significantly improves efficiency and enables breakthrough progress in NLP tasks such as translation, summarization, and question answering.

According to Apple, the new transformer model in iOS 17 can achieve automatic error correction at the sentence level. When you press the space bar, it can complete a word or an entire sentence. It also learns based on your writing style to guide its recommendations. Apple also says that speech recognition "uses a transformer-based speech recognition model that leverages a neural engine to make speech recognition more accurate."

In the keynote speech, Apple also mentioned "machine learning" many times, such as when describing the new iPad lock screen function ("When you select a Live Photo, we use an advanced machine learning model to synthesize "Extra frames"); iPadOS PDF feature ("Thanks to new machine learning models, iPadOS can recognize fields in PDFs, letting you use AutoFill to quickly fill in information like names, addresses, and names drawn from your contacts. email, etc."); AirPods Adaptive Audio feature ("With personalized volume, we use machine learning to learn how your listening preferences change over time"); and Apple Watch widget feature Smart Stack ("Smart Stack uses machine Learn to show you relevant information when you need it”).

Apple also launched a new app called Journal that uses on-device machine learning to provide personalized recommendations and bring journaling inspiration to users. These suggestions are intelligently generated based on the user's recent activities, including photos, people, places, physical training, etc., helping users start recording more easily.

Finally, during a demo of the Vision Pro headset, the company revealed that the dynamic image on the user’s eyes is a special 3D avatar created by scanning your face — thanks to machine learning, of course.

The above is the detailed content of Apple WWDC23 does not mention "artificial intelligence", preferring to use "machine learning". For more information, please follow other related articles on the PHP Chinese website!