AI Digest: Apple’s Strategic Moves in Machine Learning Take the Spotlight
Staying updated in the fast-paced realm of artificial intelligence is no small feat. While we await the day when AI can do the job for us, here’s a convenient summary of the latest happenings in machine learning from the past week, including noteworthy research and experiments that haven’t received individual coverage.
Apple made a prominent entrance into the fiercely competitive AI race last week, signaling its strong commitment to and investments in AI. At the WWDC event, the company emphasized that AI underpins many features in its upcoming hardware and software releases. Notable examples include iOS 17, which leverages computer vision to suggest recipes based on photos, and Journal, an interactive diary powered by AI that offers personalized suggestions based on user activities across various apps.
In iOS 17, Apple also introduces an enhanced autocorrect system driven by an AI model capable of accurately predicting users’ next words and phrases, including frequently used terms and even profanities. Furthermore, AI takes center stage in Apple’s Vision Pro augmented reality headset, particularly with the FaceTime feature. Through machine learning, the Vision Pro can generate a virtual avatar of the wearer, replicating intricate facial expressions down to the muscles and skin tension.
While generative AI remains a popular subcategory in the AI field, Apple’s recent moves seem aimed at reclaiming its position after facing setbacks with previous machine learning projects like Siri and its troubled self-driving car initiative. By showcasing tangible AI-infused products, Apple aims to establish its seriousness in the AI domain, countering brain drain concerns that have led talented AI scientists to seek opportunities elsewhere.
Here are other noteworthy AI headlines from the past week:
- Meta introduces MusicGen: In a bid to keep up with Google, Meta launches its own AI-powered music generator called MusicGen. The tool can convert a text description into a 12-second audio snippet and has been open-sourced.
- Regulators focus on AI safety: The U.K. government plans to host a global AI safety summit, receiving commitments from OpenAI, Google DeepMind, and Anthropic to grant early or priority access to their AI models for evaluation and safety research.
- Salesforce unveils AI Cloud: Salesforce aims to strengthen its position in the competitive AI space by launching AI Cloud, a suite of products designed to deliver enterprise-ready AI tools, as part of its cross-disciplinary efforts to incorporate AI capabilities into its product portfolio.
- Testing text-to-video AI: TechCrunch provides hands-on experience with Gen-2, Runway’s AI that generates short video clips from text. However, the technology still has a long way to go before achieving film-quality results.
- Increased funding for enterprise AI: Cohere, a startup developing an AI model ecosystem for enterprises, secures $270 million in its Series C round, indicating the availability of ample funding for generative AI startups.
- OpenAI’s stance on GPT-5: OpenAI CEO Sam Altman confirms that the organization is not currently training GPT-5, as they had previously pledged to take a pause in developing its successor, considering concerns raised by industry experts and academics about the rapid advancements in large language models.
- AI assistant for WordPress: Automattic, the company behind WordPress.com, launches an AI assistant for the popular content management system, providing users with additional support and functionality.
- Instagram’s potential chatbot: Leaked images suggest that Instagram may be developing an AI chatbot capable of answering questions and offering advice, although it remains uncertain whether these developments will be implemented.
Additionally, there are interesting developments in AI’s impact on science and research, such as a report authored by a team across six national labs exploring the potential effects of AI in these domains. Moreover, researchers at
Los Alamos are making progress in advancing memristor technology, which combines data storage and processing similar to our brain’s neurons. Natural language processing is also utilized in a study analyzing police interactions during traffic stops, identifying linguistic patterns that predict escalation, particularly with Black men.
Exciting advancements in AI’s application to healthcare include DeepBreath, a model trained to identify respiratory conditions early using recordings of patients’ breathing, and Purdue’s software that approximates hyperspectral imagery using smartphone cameras, enabling health metrics tracking without specialized hardware. MIT researchers are also inching closer to developing AI for autonomous evasive maneuvers in aviation, while Disney Research showcases a facial landmark detection network at CVPR, offering high-quality motion capture without the need for markers.
While the AI landscape evolves rapidly, staying informed about these significant developments ensures a deeper understanding of the field’s progress.
