Anthropic Introduced a new application programming interface feature (API) on Thursday to allow developers to substantiate the answers generated by Artificial Intelligence (AI) models. Nicknamed citations, the feature allows developers to restrict the generation of the Claude family from AI models to obtain documents. This aims to improve the reliability and accuracy of AI -generated answers. The AI company has already provided the resource to companies such as Thomson Reuters (for the Cocounsel platform) and Endex. Notably, the feature is available at no additional cost.
Anthropic presents a new grounding resource
Generative AI models are typically prone to errors and hallucination. This is due to the huge data sets they need to look for to find answers to the user’s queries. The addition of web searches to the equation only makes it more complicated for Large Language Models (LLMs) to avoid inaccurate information as they use relatively basic recovery generation mechanisms (RAG).
AI companies, which also create specialized tools, usually restrict data access to LLMs to improve accuracy and reliability. Some examples of such tools include twins on Google Docs, AI writing assistance tools on Samsung and Apple smartphones, and PDF analysis tools at Adobe Acrobat. However, the creation of this layer is not possible in the API, as developers create a wide range of tools that have different data requirements.
To solve this problem, the anthropic has introduced the quotation feature to its API. Detailed in a newsroom publishThe appeal allows Claude to terrify his answers to the original documents. This means that Ia Claude models can provide detailed references to the exact paragraph and the sentences that were needed to generate the output. The AI company states that this tool will make the answers generated by AI easily and more reliable.
As a result, users can add source documents to the context window, and Claude will automatically cite the source in their exit wherever they eat them from the source material. As a result, developers will not have to rely on complex warnings to ask Claude Include origin information, which the company recognized as an inconsistent and heavy method.
Anthropic claimed that with citations, developers will easily create AI solutions for document summary, tools to respond to complex consultations based on long documents as well as customer support systems.
Notably, the company stated that quotes use the anthropic’s standard token pricing model and users will not pay output tokens that return the mentioned text. However, there may be an extra charge for additional input tokens that are used to process origin documents. Currently, citations are available for the new models Claude 3.5 Sonnets and Claude 3.5 Haiku.