Merlyn Mind Launches AI-Driven Classroom Assistant Merlyn Origin

Published: November 07, 2023 at 2:42 am Updated: November 07, 2023 at 2:42 am

Edited and fact-checked:
November 07, 2023 at 2:42 am

In Brief

Merlyn Mind’s classroom AI-assistant Merlyn Origin allows teachers to use voice commands to control classroom tools, access educational content and create interactive learning experiences.

US-based AI startup, Merlyn Mind today introduced Merlyn Origin, an AI-powered classroom assistant underpinned by a large language model (LLM) designed uniquely for educational purposes, setting it apart from conventional large generalist models like OpenAI’s GPT-4.

Developed by researchers from prominent AI laboratories, including IBM Research, Amazon Alexa and Google Brain, Merlyn Mind asserts that the classroom assistant allows teachers to use voice commands to control classroom tools, access vetted educational content and create interactive learning experiences.

For example, imagine a teacher preparing for a biology lesson. With Merlyn Origin, they could say, ‘Open my presentation on biology,’ and the system would respond immediately. Need to time an activity? ‘Set a timer for 2 minutes.’ Looking to dive deeper into a topic spontaneously brought up by a student? ‘Search for the equation of gravity on WolframAlpha.’

All these actions can be executed without the teacher ever needing to turn away from the class.

“Most solutions require educators to change how they teach and adopt new applications, content, and tools, which is not the case with Merlyn, as schools have already invested heavily in modernizing classrooms with devices and applications,” Merlyn Mind’s co-founder and CEO, Satya Nitta told Metaverse Post. “With Origin, the power of LLMs and Generative AI are now accessible privately, safely, and securely in class during active instruction.”

Merlyn Mind’s Key Differentiator from Other Contemporaries

The company claims that it builds smaller, task-specific models that are compute-efficient, less expensive to train and operate.

“Multiple models can be used together in an ensemble process to ensure that the solution is private, safe, age-appropriate and accurate for teacher and student use. This is done using one model as a personally identifiable information (PII) and safety check filter, a second model as the generator of the response to the prompt and a final model as a critiquing model,” added Merlyn Mind’s Satya Nitta.

Nitta further explained that unlike ChatGPT, these LLMs do not depend on their training data to answer questions. They are instead made to use a custom corpus of content to respond to the prompts. Depending on the use case Merlyn uses a mix of open source as well as their own pre-trained models as the starting point and further finetunes them using datasets customized for the needs of education.

Tailored Content for Educational Needs

Merlyn’s models use a technique called retrieval augmented generation (RAG).

Diving deep, Nitta explained that — “In RAG, the prompt from a user is at first sent to the walled content garden (which is re-represented for ease of use by an LLM), whereupon specialized algorithms are used to search within the corpus for chunks of content that might have the appropriate response. All of these are then sorted and the most relevant of these are retrieved from the database and then presented to the LLM to generate a response.”

“From the perspective of a user, such as a teacher or a student – when they input a prompt, the system doesn’t just draw from the vast library of its initial training data to generate a response. Instead, it turns to a specific, secure content library – a “walled garden” of resources that the school or district has approved,” he added.

It also aims to ensure adaptability and safety. Schools and districts can hand-pick the content that feeds into the system, ensuring that responses are not only accurate but also aligned with educational standards, age-appropriateness, and local curricula, said Nitta.

They can populate the ‘garden’ with preferred textbooks, supplemental materials, and even district-specific documents, he added.

Further, the method ensures that the AI draws upon a vetted, relevant database to formulate its responses, thereby increasing the accuracy and reliability of the information provided. This addresses the challenge of hallucinations.

Merlyn Mind said that additional fine-tuning of the models encourages them not to hallucinate, and the inclusion of AI as a critique mechanism serves as a final check to validate the responses.

Professional Development for Educators Using AI

Nitta believes that educators are already overwhelmed with their workloads and responsibilities. Finding time to learn how to change the way they teach and the technology they teach with is difficult. However, professional development is critical to the success of any new technology in education.

“The company works closely with school customers to develop professional development programs to ensure the successful adoption and integration of Merlyn Origin into institutions and classrooms,” said Nitta. “This professional development is available as asynchronous modular training with short videos and self-serve content that educators can consume on their own time, as live training online or in-person, and as in product training where educators learn about the way to use Merlyn Origin as they are experimenting with it.”

“The company work with schools and districts to integrate into the existing formal and informal professional development infrastructure they have developed with in-house tech experts and trainers,” he added.

It deploys a train-the-trainer program to identify and train local on-site Merlyn experts who can help answer questions and drive engagement and usage of the Merlyn Origin solution.

Disclaimer

Any data, text, or other content on this page is provided as general market information and not as investment advice. Past performance is not necessarily an indicator of future results.

The Trust Project is a worldwide group of news organizations working to establish transparency standards.

Kumar is an experienced Tech Journalist with a specialization in the dynamic intersections of AI/ML, marketing technology, and emerging fields such as crypto, blockchain, and NFTs. With over 3 years of experience in the industry, Kumar has established a proven track record in crafting compelling narratives, conducting insightful interviews, and delivering comprehensive insights. Kumar’s expertise lies in producing high-impact content, including articles, reports, and research publications for prominent industry platforms. With a unique skill set that combines technical knowledge and storytelling, Kumar excels at communicating complex technological concepts to diverse audiences in a clear and engaging manner.