Digital twin
Highlights
1. Apple Intelligence is a new AI system integrated into iPhone, iPad, and Mac. Siri gets an upgrade with LLMs.
2. Zeta Labs introduced Jace, an AI agent that automates web tasks.
3. Dream Machine is a new text-to-video model that the public can now try.
4. An artificial neural network developed as a biomechanically realistic model of a rat can predict real neural patterns better than features of the rat’s movements.
5. Digital twins are revolutionizing complex scientific experiments by creating high-fidelity, real-time models of instruments.
Innovation Insights
1. How companies can use AI to find and close skills gaps (Ideas Made to Matter)
To train employees in essential digital skills, companies must accurately assess current workforce competencies and identify gaps. Artificial intelligence facilitates this by analyzing employee data to infer skill levels and areas for improvement. With many employees seeking training and willing to engage in continuous learning, AI-driven skills inference helps companies strategically plan workforce development and align training with future needs.
2. How Gen AI can make work more fulfilling (Harvard Business Review)
Research by BCG shows that high levels of repetitive, unenjoyable work (“toil”) can lead to employee turnover, while engagement with tasks that bring joy reduces this risk. In a study of administrative employees, the firm demonstrated that AI tools could decrease time spent on toil and increase time on enjoyable tasks. Crucially, the success of generative AI adoption is significantly enhanced when managers actively use and champion the technology. Effective adoption strategies include co-creating AI tool applications with employees and ensuring that AI reduces toil without diminishing the enjoyable aspects of work.
AI Innovations
1. Apple WWDC
A variety of updates were announced at Apple WWDC 2024 (The Verge):
Apple Intelligence: A new AI system integrated into iPhone, iPad, and Mac, enabling capabilities like image generation and text summarization, with a focus on security.
Enhanced Siri: Siri gets a significant AI upgrade with deeper integration and better app control, powered by large language models (LLMs).
ChatGPT Integration: Siri will now feature OpenAI’s ChatGPT, determining suitable queries for ChatGPT and processing them with user permission.
AI-Powered Apps: New AI features across iOS 18 apps, including email summarization, custom emoji creation (Genmoji), and advanced photo search in Photos.
AirPods Pro Updates: Upcoming AirPods Pro features include Siri Interactions using head gestures and Voice Isolation for improved voice clarity.
2. AI tools
LinkedIn’s new AI-powered tools for Premium members aim to enhance job searches by offering a conversational job search interface, personalized resume and cover letter assistants, and real-time guidance (LinkedIn).
Zeta Labs launched Jace, an AI agent capable of executing web tasks autonomously, and aims to automate complex browser-based tasks for consumers and businesses (VentureBeat).
Google’s novel large language model provides personalized health insights and recommendations by analyzing physiological data from wearables, achieving expert-level performance on sleep and fitness tasks through advanced contextual reasoning and code generation (Google).
3. AI models
Stability AI’s new Stable Diffusion 3 Medium offers a smaller, efficient model for image generation with only 2 billion parameters, making it suitable for consumer-grade GPUs (VentureBeat).
Luma AI launched “Dream Machine,” a platform for creating high-quality, realistic video from text prompts in minutes, offering vast potential for rapid video content creation (Luma AI).
DenseAV is a self-supervised model that learns to associate sounds with their visual sources and understand language meanings by analyzing videos, without needing any labeled data (Link).
Midjourney’s new model personalization feature tailors image generation to your preferences by learning from your interactions, requiring 200 rankings to activate, and can be triggered by adding “–p” to prompts for customized outputs (Dataconomy).
4. AI and science
Fields Medalist Terence Tao highlights that AI and proof checkers are transforming mathematics by enabling large-scale collaborations, automating tedious tasks, and potentially revolutionizing how mathematicians approach and solve complex problems (Scientific American).
To understand how the brain controls animal behavior, researchers developed a virtual rodent using a biomechanically realistic model and deep reinforcement learning, finding that its network activity predicted real neural patterns better than movement features (Nature).
5. Robotics
Stanford researchers present a system enabling humanoid robots to learn and perform tasks autonomously by mimicking human skills, utilizing reinforcement learning on human motion data, real-time shadowing for data collection, and supervised behavior cloning for task completion with egocentric vision (GitHub).
Other Innovations
1. Digital twins
Digital twins are revolutionizing complex scientific experiments by creating high-fidelity, real-time models of instruments like NASA’s James Webb Space Telescope (JWST), enabling precise monitoring and troubleshooting (MIT Technology Review).
2. Cancer vaccines
Personalized cancer vaccines, utilizing mRNA technology, show promise in preventing tumor recurrence by encoding neoantigens from an individual’s cancer, training the immune system to target residual cancer cells, as seen in ongoing clinical trials (Nature).
3. Giant wafer
Cerebras’ new wafer-scale chips outperform traditional supercomputers in molecular dynamics and AI inference by using a highly integrated design that allows for faster, more energy-efficient computations and support for unstructured sparsity (IEEE Spectrum).
