AI continues to reshape industries, governance, and education, but questions about human control remain central. A new framework argues that humans must direct science even when AI outperforms researchers. The proposed test focuses on human understanding and human direction across every phase of research, from peer review to training. It aims to keep scientific progress aligned with human values.
Meanwhile, experts warn that unchecked AI automation risks undermining fields like mathematics and science. OpenAI recently generated over 700 AI papers about Navier-Stokes equations at an unprecedented scale, highlighting concerns about quantity over depth. The Leiden Declaration promotes creativity, proof, and collaboration as essential values that AI should support rather than replace.
On the practical side, AI is finding real-world applications in unexpected places. Italian police used AI software to recover a painting missing for two decades, flagging a match within hours during a routine art fair visit in Parma. Japan's JR East railway company demonstrated a humanoid robot with a chainsaw arm to clear tree branches from rail tracks, with plans to eventually make such robots autonomous.
In education, India launched a new AI and computational thinking curriculum for students in Classes 3 to 8 starting in 2026-27. However, infrastructure gaps pose challenges: only 59.4 percent of government schools have a working computer, compared to 74.0 percent in private schools. States like Meghalaya and West Bengal report the lowest computer availability.
Product announcements reflect growing demand for AI integration in everyday workflows. Otter launched a Work mode for its AI Chat assistant that can draft emails, documents, and update Salesforce records after meetings. The feature requires human approval before completing tasks, and a study found 91 percent of business users want meeting tools to act on information from outside meetings. Separately, Coralboard introduced an edge AI development board powered by Synaptics Astra and Google's Coral NPU, giving engineers a low-power platform for prototyping offline AI products. Researchers also developed LUMINA, an AI model that turns lunar geospatial data into mission planning insights, with plans to test it using NASA's Artemis program data.
Key Takeaways
- A new framework insists humans must keep control of science, applying a two-part test focused on human understanding and human direction across all research phases
- OpenAI generated over 700 AI papers on Navier-Stokes equations, prompting experts to warn against unchecked AI automation in research
- Italian police recovered a painting missing for 20 years using AI software that flagged a match within hours at an art fair in Parma
- Japan's JR East demonstrated a humanoid robot with a chainsaw arm for clearing tree branches from railway tracks, with full autonomy planned
- India launched an AI curriculum for Classes 3-8 starting 2026-27, but only 59.4% of government schools and 74.0% of private schools have working computers
- Otter's new Work mode can draft emails, documents, and update Salesforce records after meetings, requiring human approval before executing tasks
- 91 percent of business users want meeting tools to act on information gathered outside of meetings, according to a recent study
- Coralboard offers an edge AI development board using Synaptics Astra and Google's Coral NPU for low-power, offline AI prototyping
- LUMINA AI turns lunar geospatial data into mission planning insights and will be tested with NASA's Artemis program data
- Gerol Petruzella from Williams College argues society must create new rules for governing AI as it transforms communities faster than people understand
Framework keeps human direction in science despite AI advances
A new paper argues that humans must keep control of science even when AI outperforms researchers. It proposes a two-part test focusing on human understanding and human direction. The framework applies to every phase of research including peer review and training. It aims to keep science serving human values and goals.
Understanding AI as it rapidly changes communities
An opinion piece explores how artificial intelligence is transforming society and what people need to learn about it. Gerol Petruzella from Williams College says society must create new rules for AI. The article notes that many people still do not understand what AI really is. It highlights the challenge of governing a technology that evolves so quickly.
Police use AI to find painting missing for 20 years
Italian police recovered a painting that had been missing for two decades using AI software. Officers uploaded images during a routine visit to an art fair in Parma. The AI system flagged a matching snapshot within hours. The painting showed a woman holding a half-unravelled scroll that raised suspicion among investigators.
Japan tests AI robot with chainsaw on railway tracks
Japan's JR East railway company demonstrated a humanoid robot with a chainsaw arm to clear tree branches from rail tracks. The robot is mounted on a rail truck and is currently controlled remotely by a human operator. JR East is integrating AI and robotics across its operations to improve safety and maintenance. The company plans to eventually make such robots autonomous.
India rolls out AI curriculum but many schools lack computers
India launched a new AI and computational thinking curriculum for students in Classes 3 to 8 starting in 2026-27. However, over one in three schools in India have no working computer for teaching. The gap is wider in government schools where only 59.4 percent have a working computer compared to 74.0 percent in private schools. States like Meghalaya and West Bengal have the lowest computer availability.
Otter launches Work mode for automated meeting tasks
Otter added a new Work mode to its AI Chat assistant that can draft emails, documents, and update Salesforce records after meetings. The feature is available now to all paid Otter users with no usage limits. Work mode requires human approval before completing tasks and replaces the older Express and Advanced options. A study found 91 percent of business users want meeting tools to act on information from outside meetings.
Scientists urge care in using AI for research
Experts warn that unchecked AI automation could harm mathematics and science by producing more solutions without deeper understanding. The Leiden Declaration promotes values like creativity, proof, and collaboration as essential to the discipline. OpenAI recently generated over 700 AI papers about Navier-Stokes equations at an unprecedented scale. Scientists call for AI to support rather than replace human research values.
LUMINA AI helps plan future space missions
Researchers developed an AI model called LUMINA that turns lunar geospatial data into mission planning insights. Unlike passive AI tools, LUMINA works as an active assistant alongside mission planners. The team plans to test the system on new lunar data from NASA's Artemis program. The goal is to support future human space outposts with faster and better mission preparation.
Coralboard brings edge AI prototyping to engineers
Coralboard is a new edge AI development board powered by Synaptics Astra and Google's Coral NPU. It gives engineers an open, low-power platform for prototyping AI products that work without internet. The board is available from DigiKey and supports hardware-accelerated Gemma 3 270M. It aims to help developers move quickly from experimentation to production-ready edge AI devices.
Sources
- Why and How to Do Science When AI Is Better at It: A Framework Grounded in Human Understanding and Direction
- AI is changing our world. First, let's understand what it is.
- This painting went missing 20 years ago. Police tracked it down with the help of AI
- Japan experiments with autonomous AI chainsaw-wielding robots — video shows rail truck-mounted metal humanoid cutting down tree branches
- AI curriculum is being rolled out. But where are the working computers?
- Otter adds ‘Work’ mode to automate more meeting-related tasks
- Why (and How to) Do Science in an Age of AI
- Could next-gen artificial intelligence help map out future space outposts?
- New Edge AI Development Board for Faster, Private Prototyping
Comments
Please log in to post a comment.