Scientists around the world are getting new AI helpers that can work like research partners. These smart systems, called agentic AI, don't just answer questions. They can think, plan, and do research tasks on their own.

The United States National Institutes of Health (NIH) made big news this week with a new AI tool called GeneAgent. This smart helper studies genes, which are like instruction books inside our bodies. GeneAgent can look at thousands of genes at once and figure out what jobs they do. This helps scientists understand diseases better and find new medicines.

What makes GeneAgent special is that it can check its own work. Many AI systems sometimes give wrong answers, called "hallucinations." But GeneAgent has a built-in checker that compares its answers with what scientists already know. If something doesn't match, it tells the researchers. This makes the tool much more trustworthy for important health research.

At Carnegie Mellon University in the United States, scientists got $10.5 million from the government to start a new AI research center. This center will teach AI systems to be math detectives. The AI will try to solve hard math problems, create new math theories, and even draw pictures to help explain complex ideas. This is exciting because math is the language that helps us understand everything from space travel to computer games.

The math AI center wants to connect two different types of AI thinking. Some AI is good with symbols and rules, like a calculator. Other AI is good with patterns, like recognizing faces in photos. By combining both types, the new center hopes to create AI that can discover new math concepts that humans haven't thought of yet.

AI battery research made headlines when scientists used smart computers to design new materials for batteries. Regular battery research can take many years because scientists have to test thousands of different materials. But AI can simulate these tests in computers, making the process much faster. The new materials could make electric car batteries last longer and charge in minutes instead of hours.

Safety got a boost this week with new deepfake detection technology. Deepfakes are fake videos that look real, showing people saying or doing things they never actually did. A new AI system can spot these fake videos with 98% accuracy. This is important because fake videos can trick people and spread false information. The new detector works on different types of fake videos, whether they change someone's face or voice.

Companies are also jumping into agentic AI. Thomson Reuters, a big information company, launched CoCounsel Legal for lawyers. This AI helper can do legal research, write documents, and handle complicated legal tasks. Instead of lawyers spending hours reading through law books, the AI can do this work and give them the important information they need.

The shift to agentic AI represents a big change in how we think about artificial intelligence. Old AI systems waited for humans to ask questions and then gave answers. New agentic AI systems can set their own goals, make plans, and work toward solving problems without constant human guidance. They can even work together with other AI systems or human researchers as a team.

Experts say this change is as big as when the internet first started. Just like the internet changed how we get information, agentic AI will change how we do research and solve problems. These AI helpers won't replace human scientists, but they will make research faster and help us discover new things we never could have found on our own.

The trust and safety of these AI systems is very important. Scientists are working hard to make sure AI agents can explain their decisions and be reliable. This is especially important in areas like medicine and science, where wrong answers could be dangerous. Researchers are building special monitoring systems to keep track of what AI agents do and make sure they're working correctly.