Librari Help Center
Librari isn’t connected to the internet, and currently has limited knowledge of world events beyond 2021, so it’s best not to use Librari to find anything current, like news, the very latest books, sports results or stock information.
In the future, Librari will have more recent information, but for now don’t use it for anything current.
For every “action” we provide numerous (4,500,000 in total) curated examples. You can click any example and see the results, or you can use the examples as “templates” for your own queries. Remember that the examples are only examples, meant to guide you, and you are free to make whatever queries you like (though we recommend phrasing them in a way similar to what you see in the examples in order to get the best results).
Librari delivers simple answers in an accessible way, in one spot, and doesn’t ask you to rifle through blue links (or to risk malware or spyware by visiting external websites).
In addition, Librari delivers detailed answers, as well as creative answers, in a way that search engines simply cannot.
1. Curated prompt examples: Librari uses a collection of 4,500,000 meticulously crafted and tested human-engineered prompts designed to guide the AI in producing reliable, thoughtful, and accurate responses.
2. Prompt engineering: Expert prompt engineers carefully create prompts to request specific, fact-based, and unbiased information. By framing queries with clarity and precision, the AI is less likely to generate hallucinated or fabricated responses.
3. Model settings: Adjusting the parameters of the AI model can influence the output quality and restrain potential hallucinations. Settings that Librari use produce more focused and less random outputs, helping to mitigate the risk of generating made-up content.
4. Continuous improvement: AI models, including Librari, benefit from ongoing development and user feedback, which help identify inaccuracies or instances where the AI generates undesired content. This iterative process promotes improvements to the system, reducing the chances of hallucination over time.
Although these measures help minimize the risk of hallucination, it is crucial for users to remain vigilant and aware of the potential limitations associated with AI-generated information. If in doubt, users should always verify the information through other trusted sources or seek expert advice to ensure accuracy.
Librari is a tool made specifically for libraries, so that patrons like you can have a world of great information at your fingertips for free, without having to visit external websites or rely upon search engines.
You can ask Librari questions using your computer’s microphone. You can even ask questions in dozens of languages and it will understand you.
Fun tip: Use the microphone to have chats in different languages with Librari’s multilingual bots!
You can toggle between text mode, which allows you to submit textual queries, and image mode, where you can ask Librari to generate images for you.
Yes, after doing a search you can hover your mouse over the results area and you’ll see a “copy” icon. Click that to copy the contents to your clipboard and paste it into another application of your choice.
Librari improves and evolves through a combination of technological advancements, user feedback, prompt engineering, and continuous learning. Here’s an overview of each aspect:
1. Technological advancements: As the technology behind AI models like OpenAI’s GPT series advances, these improvements are incorporated into Librari. Upgrading to newer and more powerful AI models enables Librari to provide better-quality responses, increased knowledge depth, and higher accuracy.
2. User feedback: User feedback plays a critical role in Librari’s evolution. By evaluating the AI-generated responses based on their accuracy, relevance, and comprehensiveness, both users and developers can identify areas that need improvement. Feedback loops help refine the system’s understanding and responses to various topics and user inputs.
3. Prompt engineering: Continual improvements in prompt engineering enhance the ability to ask questions more effectively, obtaining better-quality and more targeted responses from Librari. Expert engineers and prompt developers constantly refine existing prompts and create new ones to cover different topics and use cases.
4. Continuous learning: Although Librari primarily utilizes its training data up to a certain point, OpenAI frequently trains new AI models and updates existing ones. By incorporating more recent and diverse data sources, Librari can keep updating its knowledge, becoming more accurate and relevant to users over time.
These factors work together to ensure that Librari evolves and improves in providing accurate, helpful, and reliable information to users while refining its interactions with them and adapting to new areas of knowledge.
Robust content moderation, and curated prompt engineering, are built into the system, making it difficult to use the system for purposes other than what it was designed to be used for.