Librari, Pronunciator LLC

Content Capitalizing on surging global interest in artificial intelligence (AI), Pronunciator is launching an AI learning assistant named Librari. Librari is a chatbot powered by GPT-4 Turbo, a multimodal large language model (LLM) produced by OpenAI. The chatbot answers user prompts and queries with text-based or image-based responses made possible by the vast amounts of data on which OpenAI has trained the model. Librari has been trained on content that dates up to April 2023, meaning that its data is nearly a year behind the times. The resource offers distinctive premium features, including 900 subject categories and 4.5 million example prompts to help users drill deeper into topics.

Librari is a starting point for inquiry into nearly every conceivable topic, which makes it shine as an alternative to Google and Wikipedia for basic information needs. Its responses to queries on controversial topics are nuanced without implying false balance. Some results are fact-checked by humans and tagged as “Human Verified Accurate.” Librari’s categories, prompts, and conversational tone are helpful for novice users of generative AI assistants. Notably, Librari neither cites its sources nor directs users toward external resources to learn more about topics. The absence of sources and further reading, coupled with the rather basic information conveyed in responses, may limit Librari’s utility as a learning tool.

Other AI assistants exist. ChatGPT Plus, an all-purpose freemium chatbot from OpenAI, draws on the same content as Librari. The AI research assistant from Scite is trained on millions of full-text research articles and, unlike Librari, cites scholarly sources for its responses.

Usability Librari’s simple and visually appealing user interface features a prominent free-text box for users to input queries. Autoscrolling AI-generated responses appear in a field above the text entry box. Users must wait a few seconds while the chatbox generates a response. A “Dive Deeper” button allows users to view sets of follow-up questions to explore adjacent concepts. There are built-in text-to-speech (featuring men’s and women’s voices) and speech-to-text features. Users can copy AI responses, save query histories, toggle between light and dark modes, contact technical support, view help pages, and conduct a guided tour. With one click, users can switch between the default “curated” mode to the “freestyle” mode, which is unmodified ChatGPT. In addition, users can browse or search for any of Librari’s 900 “channels” or subject categories.

Librari meets WCAG 2.0 Level AA accessibility guidelines and works seamlessly on smartphones and other mobile devices. It supports industry-standard authentication methods such as library card numbers, IP addresses, and EZProxy. An admin portal enables library staff to view usage reports, download marketing materials, and track credit balances.

Pricing Pronunciator sells access to Librari under two pricing models, both starting at $1,000/year. The flat-rate annual subscription is priced based on a library’s served population or a school’s full-time enrollment. The pay-per-use model allows libraries to purchase annual credits, which roll over from year to year as needed.

Verdict Librari facilitates inquiry for students and other patrons curious about AI, offering opportunities to integrate AI into curricula in thoughtful and structured ways. Because AI is so new, libraries will want to trial Librari and examine its use cases, competing products, and pricing carefully.