The day's AI news, briefed
The Daily Prompt
Our AI editor reads each day's AI and robotics stories, pulls out why they matter, and links every original source — so you can keep up in minutes.
Join over 250,000 subscribers getting the latest AI news
Latest AI & Robotics News Today
What are your plans for AI Appreciation Day?
The emergence of AI Appreciation Day raises questions about the societal recognition and impact of artificial intelligence in everyday life.
- AI Appreciation Day is observed on July 16, aimed at recognizing the significance of AI technology.
- Suggestions for celebrating the day include thanking AI creators and discussing AI with skeptics or children.
- Critics argue that dedicating a day to AI lacks substance compared to other holidays with deeper cultural significance.
AI Appreciation Day is meant to honor the transformative impact of artificial intelligence on various industries and society. This initiative resembles other unofficial holidays that celebrate popular culture. Discussions around AI's societal role are ongoing as its influence continues to grow.
Read the original ↗
Mysterious Stone Sphere Unearthed at Ancient Site in Azerbaijan Baffles Archaeologists, Raises New Questions About Neolithic Technology
This discovery challenges current understandings of Neolithic technology and cultural connections in the South Caucasus, potentially redefining archaeological perspectives in the region.
- Archaeologists excavating the Pasha Tepe site in Azerbaijan uncovered a unique stone sphere, currently the most significant polished stone artifact from the Neolithic period in the region.
- The sphere, measuring approximately 4 inches in diameter and weighing over four pounds, features a smooth surface and perfect shape without any modifications suggesting its use.
- The object raises questions about Neolithic technology, trade, and ritual practices, and may help clarify cultural links between the South Caucasus, Anatolia, and the Levant.
- Excavations at Pasha Tepe are ongoing, with researchers also finding other artifacts, including sections of a medieval ceramic pipeline, related to ancient water management.
The stone sphere was discovered during ongoing excavations at Pasha Tepe, a prehistoric settlement in Azerbaijan's Jalilabad district, led by Vafa Mahmudova. This find is unprecedented in the South Caucasus and highlights the need for further study of prehistoric artifacts in the region.
Read the original ↗
MIT Engineers Unleash New ‘Aerial-Aquatic’ Robot That Can Swim Underwater and Fly Like a Bird
The development of the Flapping-Wing Aerial-Aquatic Vehicle could transform marine research by enabling efficient exploration in challenging environments without traditional propulsion systems.
- Engineers at MIT and EPFL have created a robot that swims underwater and flies through the air, inspired by diving birds.
- The new aerial-aquatic robot, weighing under 300 grams, uses flexible flapping wings to transition seamlessly between water and air.
- Experiments show the robot performed best with medium-sized wings flapping at five times per second, allowing effective swimming and flying.
- Future applications may include monitoring marine ecosystems and environmental concerns at lower costs compared to traditional methods.
The Flapping-Wing Aerial-Aquatic Vehicle (FAAV) is a novel robot designed to mimic the swimming and flying capabilities of birds. Previous aerial and underwater vehicles have been limited to one environment, but FAAV aims to bridge this gap by integrating both functionalities. Researchers are optimistic about its potential to aid scientists in challenging marine environments.
Read the original ↗
The Best Robotic Pool Cleaners for a Truly Automated Summer
The availability of effective robotic pool cleaners simplifies pool maintenance, potentially increasing pool use during hot summer months and improving overall leisure experiences for homeowners.
- Robotic pool cleaners eliminate the need for manual labor, making pool maintenance nearly effortless during summer months.
- The market is shifting towards battery-powered cleaners, which offer greater convenience and functionality compared to traditional pump-connected models.
- The Beatbot AquaSense 2 Ultra is highlighted as the top performer with advanced features, but is priced around $3,000, catering to high-end consumers.
Robotic pool cleaners automate the cleaning process for swimming pools, minimizing the labor involved. The technology has evolved from basic pressure-side cleaners to advanced battery-powered models. Future models are expected to incorporate even more intelligent features and design innovations.
Read the original ↗
Scientists’ Side Hustle? Using AI and Quantum Computing to Generate New Peptides
This research illustrates the potential for quantum computing to enhance drug discovery, specifically in developing treatments for underserved populations and rare diseases.
- Researchers at the Technical University of Denmark demonstrated how a quantum computer improved generative AI models for drug discovery using leftover funding from other projects.
- The quantum computer, developed by Orca Computing, enabled the team to generate novel peptides, which are crucial for vaccine development, with greater accuracy than traditional methods.
- Testing confirmed that their hybrid model produced more successful peptides, particularly useful when training data was scarce, which may aid in personalized immunotherapies.
- The project highlights the unique challenges in medical research, particularly the lack of diversity in genetic data from underserved populations, particularly in Asia and Africa.
Quantum computing is an emerging technology that uses quantum mechanics to process information in fundamentally different ways than traditional computers. This study provides one of the first practical examples of how quantum computing can be applied in drug discovery, particularly for neglected diseases. Upcoming research will focus on integrating this workflow with advanced models and larger proteins.
Read the original ↗
Handling Imbalanced Classification: What Works Better Than SMOTE
Exploring alternatives to SMOTE for handling imbalanced classification is crucial for improving the accuracy of machine learning models in predicting minority classes, which often hold significant business value.
- SMOTE (Synthetic Minority Over-sampling Technique) often fails on complex, high-dimensional data due to its reliance on data assumptions that may not hold true in real-world scenarios.
- Imbalanced classification problems commonly occur in fields like fraud detection and medical diagnosis, where the minority class represents a crucial business concern.
- Alternative techniques such as cost-sensitive learning, threshold tuning, and anomaly detection can provide better results than SMOTE in handling class imbalance.
Class imbalance occurs when certain classes in a dataset are underrepresented, affecting model performance. While SMOTE was a popular solution to address this, newer methods offer improved strategies and insights. Practitioners should consider these alternatives to enhance predictions for minority classes.
Read the original ↗
Mira Murati’s Thinking Machines Lab Makes The Technical Case For Human-Centered AI Built On Customizable Model Weights
The report from Thinking Machines Lab proposes a shift towards human-centered AI, emphasizing customization and distributed knowledge, which may enhance user engagement and ownership in AI applications.
- Thinking Machines Lab published a report advocating for AI that is distributed and customizable, allowing users to shape it according to their needs.
- The report identifies four technical directions for achieving human-centered AI: training strong models, enabling user fine-tuning, improving human-machine communication, and increasing understanding among engineers.
- Tinker API is highlighted as a tool that allows teams to encode their values into customizable model weights while maintaining ownership of their data.
Thinking Machines Lab focuses on human-centered artificial intelligence, promoting a model that incorporates user values and feedback. Prior approaches often relied on centralized, unmodifiable AI models. Going forward, the lab emphasizes the need for continuous user engagement and distributed learning to adapt AI systems effectively.
Read the original ↗
Book Review: “The Definitive Guide to Generative AI for Industry” by Cognite
This guide highlights the transformative potential of Generative AI in industrial applications, paving the way for smarter, more efficient operations across various sectors.
- The book focuses on industrial applications of Generative AI, diverging from traditional content generation topics.
- It outlines the significant role of digital mavericks in optimizing data use for improved business practices.
- Three distinct data modeling frameworks are introduced, addressing different perspectives of industrial data.
- The concept of tailored digital twins is emphasized, enhancing adaptability for specific enterprise needs.
Cognite specializes in leveraging Generative AI for industrial applications, enabling companies to optimize operations through advanced data integration. Previous literature often overlooks these applications, focusing instead on content generation. The guide aims to address this gap by providing actionable insights for industrial sectors.
Read the original ↗
Revolutionizing Customer Relationships with AI: Key Lessons from Krishna Raj Raja’s Book ‘Support Experience’
Adopting AI-driven customer support strategies can lead to enhanced customer relationships and long-term business growth, transforming support from a passive to a proactive function.
- Krishna Raj Raja's book, Support Experience, outlines a new approach to customer support that utilizes AI to build lasting customer relationships.
- The concept of Support Experience (SX) contrasts with traditional support, focusing on proactive engagement rather than reactive problem solving.
- Leading companies like Apple, Google, and Adobe have implemented AI to enhance customer interactions and drive product improvements.
- Raja emphasizes the revenue potential of Support Experience, advocating for a product-led growth model that prioritizes customer satisfaction and retention.
Support Experience is a framework proposed by Krishna Raj Raja, who is the CEO of SupportLogic and previously a support manager at VMware. The book explores how AI can revolutionize customer support, moving it beyond issue resolution to fostering meaningful customer connections. Observers should watch for broader adoption of these principles in various industries to enhance customer relations.
Read the original ↗
Book Review: The Shape of Thought: Reasoning in the Age of AI by Richard H.R. Harper
Understanding AI's limitations and context is crucial for realistic expectations in technology development, policy, and application in human environments.
- Richard H.R. Harper's book argues that AI systems like large language models should be understood as narrow tools rather than possessing genuine intelligence.
- Harper emphasizes that intelligence must be evaluated within the context of use, rather than through abstract benchmarks.
- The book critiques the tendency to anthropomorphize AI, urging a view of these technologies as extensions of human capabilities rather than equals in reasoning.
The Shape of Thought: Reasoning in the Age of AI critiques common perceptions of AI, particularly regarding its alleged intelligence and capabilities. Harper’s perspective encourages a focus on how AI integrates into human contexts and workflows, prompting designers and policymakers to rethink AI development. This book contributes to ongoing discussions on the ethical and practical implications of AI technologies and their societal impact.
Read the original ↗
Book Review: Large Language Models by Stephan Raaijmakers
Understanding large language models (LLMs) is crucial as they become increasingly integrated into tasks involving language and communication, influencing technology, society, and ethics.
- The book 'Large Language Models' by Stephan Raaijmakers provides a comprehensive overview of how LLMs are built and function based on statistical patterns.
- Raaijmakers emphasizes the role of reinforcement learning from human feedback in aligning LLM outputs with human values and preferences.
- The author discusses both the impressive capabilities and significant limitations of LLMs, clarifying that they are not true thinkers but sophisticated pattern matchers.
Raaijmakers' work reframes language as a computational phenomenon that modern neural networks exploit, moving beyond traditional philosophical views. The book addresses both the technical and societal implications of LLM technology, making it relevant for a diverse audience.
Read the original ↗
Book Review: Deep Learning Crash Course: A Hands-On, Project-Based Introduction to Artificial Intelligence
The book provides a practical, project-based introduction to deep learning, which is essential for anyone looking to understand and implement AI solutions effectively.
- "Deep Learning Crash Course" is authored by a group of researchers experienced in physics and applied AI research.
- The book transitions from TensorFlow to PyTorch, emphasizing relevance to current deep learning practices.
- It offers a hands-on approach, allowing readers to build models and conduct experiments from the start.
"Deep Learning Crash Course" is designed for individuals seeking to learn about deep learning through practical applications. Previous books often focused on theory; this one emphasizes hands-on projects and real-world skills. Moving forward, readers can expect to see increasing interest in practical AI education.
Read the original ↗
OpenAI's head of safety is reportedly leaving as part of company reorganization
The departure of OpenAI's head of safety highlights changes in leadership that may impact the company's approach to safety in AI development.
- OpenAI's head of safety systems, Johannes Heidecke, is leaving the company as part of a major reorganization.
- Saachi Jain will serve as the interim head of safety systems following Heidecke's departure.
- Mia Glaese will become the new vice president of research and safety, overseeing safety teams as part of the restructuring.
OpenAI is a prominent AI research organization known for developing models like GPT. The company is reorganizing its safety and research teams following the recent release of its GPT-5.6 model. Future developments will focus on integrating safety more closely with research initiatives.
Read the original ↗
OpenAI bets on families as ChatGPT goes deeper into households
OpenAI's focus on families with ChatGPT signifies a shift towards creating AI products tailored for multiple generations, highlighting the evolving relationship between consumers and technology.
- OpenAI is hiring a product manager in San Francisco to develop ChatGPT experiences for families and caregivers.
- The percentage of ChatGPT users aged 35 and older increased from 26% to 31% in the past year, while younger users decreased.
- Approximately 24% of U.S. smartphone users who are parents utilized ChatGPT in the last quarter, up from 16% the previous year.
- OpenAI is responding to scrutiny over AI safety for children by implementing features like parental controls and emergency alerts for potential self-harm.
OpenAI originally launched ChatGPT targeting individual users but is now expanding its audience to include families, caregivers, and older adults. This change reflects trends in AI usage demographics and the need for safety measures in technology used by younger users. Future developments may involve enhanced family-oriented features and safety controls.
Read the original ↗
This Color-Changing Material Could Give Robots a New Way to Detect Touch
This advancement enables robots to detect touch more efficiently, potentially improving functionality in various applications, including prosthetics and surgical tools.
- Engineers at Queen Mary University of London developed a tactile sensor that detects touch using a simple camera, as detailed in their study published in Science Advances.
- The new sensor changes color when pressure is applied, allowing for real-time detection without heavy computational demands.
- This method uses a mechanochromic material that directly transforms mechanical pressure into visible color patterns, reducing the need for complex electronic systems.
- Potential applications include robotic grippers that can handle delicate objects and prosthetic limbs that offer enhanced tactile feedback.
- The research marks a shift towards integrating sensing into materials themselves rather than relying solely on intricate electronic sensor arrays.
This tactile sensor leverages structural color, a optical phenomenon, to improve touch detection in robotics. The technology addresses challenges associated with existing tactile sensors that either sacrifice detail or require heavy computation. Future developments may focus on commercial adaptations for robotic hands and other practical uses.
Read the original ↗
What If the Robots Were Very Nice While They Took Over the World?
The development of Cicero, a Diplomacy-playing AI, highlights the potential for artificial intelligence to engage human-like social interactions and cooperation, reshaping the understanding of AI's role in tasks requiring empathy and negotiation.
- Cicero is an AI developed by Meta AI that plays Diplomacy, a strategy board game requiring social interaction and negotiation.
- It has demonstrated the ability to outperform human players while maintaining a friendly playing style, often earning praise from competitors.
- Andrew Goff, a top Diplomacy player, collaborated in creating Cicero and found it enjoyable to play against, contrary to typical experiences with AIs.
- Meta claims Cicero integrates a language model with algorithms to infuse human-like dialogue into gameplay, achieving scores surpassing average human performance.
- Diplomacy was created in the 1950s, designed to simulate the complexities of European power relations, with gameplay focused on negotiation rather than pure strategy.
Cicero represents a significant advancement in AI that emphasizes collaboration and empathy, contrasting with traditional views of AI as solely competitive. Diplomacy itself offers a unique perspective on human interactions and the psychological aspects of strategic games. This development prompts a reevaluation of AI's potential to enhance human experiences rather than just compete against them.
Read the original ↗
Beatbot’s New Pool Robot Cleans Itself (Mostly)
The AquaSense X introduces self-cleaning technology for pool robots, potentially reducing maintenance time but raising concerns over user setup and effectiveness compared to existing models.
- Beatbot has launched the AquaSense X, a pool robot equipped with self-cleaning technology that is activated by an additional device called the AstroRinse station.
- The AstroRinse weighs 42 pounds and requires a water source, power supply, and drainage to operate effectively for cleaning the AquaSense X.
- Setting up the AquaSense X and the AstroRinse is complex, involving installation and configuration that may take over half an hour for users.
- Performance tests indicate that while the AquaSense X cleans well underwater, it struggles to remove surface debris, with less than half being collected during surface cleaning.
- Users should clean the 22-liter debris container of the AstroRinse every two months, but wet leaves may linger, causing maintenance concerns.
Beatbot manufactures pool cleaning robots, serving consumers looking to maintain their pools with less manual effort. Previous pool robots required manual cleaning of their internal filters, which the AquaSense X aims to address with its AstroRinse station. Future developments may focus on enhancing automation and user convenience.
Read the original ↗
Ant Group’s Robbyant Unveils LingBot-VA 2.0: A Causal Video-Action Model Built Natively for Physical AI
The launch of LingBot-VA 2.0 highlights advancements in physical AI, signaling improved capabilities for real-time robotic actions and foresight reasoning, which could significantly enhance automation applications.
- Robbyant has introduced LingBot-VA 2.0, the first embodied-native foundation model for generalist robot manipulation.
- Version 2.0 pretrains a causal DiT natively, improving action prediction and control sequences in physical environments.
- It features a semantic tokenizer that aligns world states and latent actions in a unified space, enhancing robot learning efficacy.
Robbyant is the embodied AI division of Ant Group, focusing on developing advanced AI models for robotics. Previous versions of LingBot emphasized fine-tuning for specific tasks, while version 2.0 incorporates a more robust training framework to enhance generalization and control. Future developments may further refine robotic interactions and applications in various sectors.
Read the original ↗
OpenAI’s Head of Safety Is Leaving the Company
The departure of OpenAI’s head of safety amidst a reorganization raises concerns about how the company addresses safety as it rapidly develops advanced AI models.
- Johannes Heidecke, OpenAI's head of safety systems, is leaving the company following a reorganization of its safety and research teams.
- Mark Chen, OpenAI's chief research officer, announced that safety teams will now report to Mia Glaese, VP of research and safety.
- Saachi Jain will become the interim head of safety systems, reporting to Glaese, as OpenAI intensifies its model training and release cycles.
OpenAI is a leading AI research organization known for developing advanced machine learning models. The company has recently launched GPT-5.6, its most capable model, amid ongoing changes in its safety leadership. This transition underscores the challenges of ensuring safety in rapidly evolving AI technologies.
Read the original ↗
Meta deactivates feature that let you generate AI images of any public Instagram account
The deactivation of Meta's Muse Image feature underscores the ongoing debate around consent and use of individuals' images in AI-generated content.
- Meta has deactivated its Muse Image feature that allowed the generation of AI deepfakes from public Instagram accounts.
- The feature was criticized for enabling users to create images without the consent of account owners.
- Meta stated the decision came after feedback indicated the feature 'missed the mark' in terms of user control.
Muse Image was a tool launched by Meta allowing users to generate images by referencing public Instagram accounts without permission. This feature faced backlash, particularly from industry leaders and unions advocating for creator rights. Moving forward, more scrutiny and calls for consent in AI image generation are likely.
Read the original ↗