Imagine a search engine that doesn't just give you links, but actually understands your questions and gives you direct answers. And what if it could prove where it got that information from? This isn't science fiction anymore. A new tool is changing how we find things online, offering a smarter, more reliable way to get the answers we need.

This new system uses advanced AI, similar to the technology behind tools like ChatGPT, but with a crucial difference. It can look at the live internet and pull information directly from websites to answer your questions. Think of it as a super-smart assistant that can do its own research and tell you exactly where it found each piece of information.

How This AI Search Works

The brain behind this system is a special AI model. It's been trained using a massive amount of search data, learning from how people look for information and what kind of answers are most helpful. This training helps the AI understand complex questions and find the right web pages to gather information.

Unlike older AI models that relied only on the data they were trained on, this new AI has the ability to browse the live web. This means it can access the most current information available. It's like giving the AI a pair of eyes to see the internet as it is right now, not just a snapshot from the past.

The

Power of Cited Sources

One of the biggest problems with current AI answers is that they can sometimes make things up or present information as fact when it's not. This new tool tackles that issue head-on. When the AI gives you an answer, you can see exactly which website it came from.

This ability to *cite sources

• is a game changer. It builds trust because you can check the information yourself. If you see an answer, you can simply hover over it or click on the text to be taken directly to the original webpage. This makes the AI much more reliable for serious research.

"Giving the model web access lessens its burden to need to store a snapshot of human knowledge within its parameters. Rather, it knows how to piece together primary sources in a natural and informative way."

This approach means the AI doesn't have to try and memorize everything. Instead, it learns how to find and combine information from different places, much like a human researcher would. This makes the answers more natural and easier to understand.

Comparing to Other AI Tools

Many popular AI tools work by using a vast amount of data they were trained on. This data is like a giant library, but it can become outdated. If you ask about something that happened yesterday, the AI might not know.

This new AI, however, can go out and find that information. It's trained to find the best, most up-to-date sources and use them. This makes it incredibly useful for topics that change quickly or require the very latest details.

Another benefit is cost. Building and running large AI models can be very expensive. This new method, which uses web browsing, is *significantly cheaper

• to operate. This could lead to more advanced AI tools becoming available to everyone.

Limitations and Future Possibilities

This AI isn't perfect, and its creators are upfront about its current limits. Because it relies on finding good sources on the web, it might not always provide a complete answer if there isn't enough reliable information out there.

Sometimes, AI models can be very creative and generate answers even when they aren't sure. These creative answers can be fun, but they might also contain errors. This new AI is designed to be less creative in those situations and stick to what it can verify.

The team plans to show both types of answers in the future. You might see answers that are directly supported by sources, and perhaps other answers that are more speculative or creative, clearly marked as such. This gives users more choice and control over the information they receive.

Real-World Examples

To show how this works, here are a few examples of questions you could ask and the kind of results you might get:

• *Setting a cookie in a web development framework:

• The AI could search for documentation and tutorials on how to do this in a specific language like FastAPI, providing code examples and links to the official guides.

• *Learning from users:

• If you asked about what a famous tech figure learned from users, the AI could find interviews or articles where they discussed their experiences and provide direct quotes.

• *Getting command-line parameters in a programming language:

• For a language like Rust, the AI could find guides on how to access arguments passed to a program when it runs, offering code snippets.

• *Understanding business decisions:

• Asking why a prominent figure bought a major social media company could lead the AI to find news reports, official statements, and analyses explaining the motivations behind the purchase.

These examples show the AI's ability to handle a wide range of topics, from technical how-tos to understanding complex events.

The

Future of Online Search

Tools like this represent a significant step forward. They combine the conversational power of AI with the reliability of real-world information. The ability to check sources means we can trust the answers we get more than ever before.

As AI continues to develop, we can expect even more sophisticated tools. Search engines might become less about finding links and more about providing direct, verified knowledge. This could change how we learn, work, and interact with information online.

This new approach is making AI search more useful and trustworthy. By focusing on verifiable information and clear sourcing, it's setting a new standard for how we find answers on the internet.