Have you heard of Loki? No, I’m not referring to the mischievous god from Norse mythology, but rather a tool that might revolutionize the way we track fake news. Loki is an open-source solution designed to sift through long texts, extract dubious claims, and verify them in the blink of an eye.
Loki is an extremely promising project for anyone passionate about fact-checking, whether you’re a journalist, researcher, or just a concerned citizen aiming to distinguish truth from falsehood. Its main strength lies in its comprehensive pipeline, which handles the entire process from breaking down texts into individual claims to the final verification.
So, how does it work in practice?
Loki starts by breaking down a lengthy text into smaller, more manageable pieces to clarify the content. Then, like a meticulous detective, it evaluates each claim to determine if it warrants verification. Vague or ambiguous statements are set aside; the focus is on concrete details!
Once the interesting claims are identified, Loki transforms into a web detective. It generates precise queries to search for evidence on the Internet, using various AI models such as natural language processing (NLP), machine learning (ML), and information extraction. This means you no longer have to spend hours sifting through sources; Loki handles it all!
After gathering the evidence, Loki takes on the role of investigator to assess the validity of each claim. Like an impartial judge, it examines the evidence for and against to render its verdict. Fake news or reliable information? You’ll find out in an instant! However, it’s important to remember that even though Loki relies on advanced AI models, no tool is perfect. Human supervision is still necessary to ensure the reliability of the results.
The interface is basic but allows direct interaction with the verification engine. You don’t need to be a tech expert; it’s accessible to everyone.
What’s great is that Loki doesn’t just tell you whether information is true or false. It also explains how it reached that conclusion, with the transparency you would expect from such tools. This allows for sourcing and justifying the analysis. Additionally, it’s educational, as it not only uncovers fake news but also provides a corrected and sourced version of the information. This makes it possible to become a fact-checking pro in no time!
If you want to try Loki yourself, it’s simple. Just go to the site https://loki.librai.tech/ to explore all the features of this tool. Meanwhile, here’s a quick tutorial to get started with Loki. You’ll need API keys for OpenAI, Anthropic, and Serper.
First step: Install Python and Poetry by following the instructions on the official site.
Then, clone Loki’s GitHub repository and start the installation with the following commands:
git clone https://github.com/Libr-AI/OpenFactVerification
cd OpenFactVerification
pip install -r requirements.txt
poetry install
It might seem a bit technical, but it’s manageable for everyone!
Once everything is set up, you can start experimenting with Loki.
To verify a text, it’s straightforward:
python -m factcheck --modal text --input mon_texte.txtLoki will get to work and provide a complete analysis of the text, ending with a detailed JSON report.
You can also test with different types of content (image, video, audio…) by changing the --modal option. And if you want to customize the model used or the sources of verification, simply pass a configuration file as a parameter:
python -m factcheck --modal text --input mon_texte.txt --api_config ma_config.yamlNote that all of this is available as a Python library, making it easily integrable with your tools.
With this, you have all the tools you need to become a fact-checking expert. So, give it a try! And if you need help, remember that the Loki community is available on GitHub or on the project’s dedicated Discord server.
