Generative AI Usage in Nualang

How generative AI is used in Nualang to help teachers get more done, more quickly.

Generative AI is a powerful set of new tools and applications that is revolutionizing how we create and interact with digital content. At Nualang, we are using some of these tools to help us in various parts of our own business and bringing some new features to our products that make use of these tools for the benefit of our customers.

In this blog post, we will introduce you to some of the ways in which we are using Generative AI at Wayside and in Nualang.

As we take a tour through some of these features, it is helpful to have some basic understanding of how generative AI works and to help with that, you might want to refer to one of our previous blog posts Generative AI - How it Works.

Privacy and Proprietary Data

Some of the concerns that are often raised when dealing with any third party services and particularly with AI and machine learning are:-

• How is the safety and privacy of personal information managed?
• Is information passed to subsidiary service providers and how is it managed by them?
• Is proprietary and or private information used to train publicly accessible models?

Wayside respects data privacy and personal information and manages it appropriately. You will find some more technical details below on how we manage data in the context of AI services and AI models.

Nualang AI Usage

AI is used in a number of Nualang's services to help users to get more done, more quickly. At this point in time, such services are only available to our own content creators and teachers. Students do not have direct access to AI services. Below, we outline the features of Nualang that offer generative AI aids.

Content Generation

In various parts of Nualang, teachers have always been able to create courses and course materials. Now, AI services are optionally available to help with that task.

• Phrases - Teachers can ask AI to generate phrases. The phrase generation is based on the general content of the course and is appropriate to the level of the class.

• Roleplays - Similarly, teachers can ask the AI to generate roleplays and again the context of course content and class level is used. In the case of roleplay, there is a further option to have generative AI help with the generation of Comprehension Questions inside the roleplay editor.

• Chatbots - Again, the AI can help the teacher to generate a bot, using the context of course content and class level.

In all cases the teacher has the option to accept or decline to use the material produced by the AI or alter it. Or, the teacher can opt not to use the AI generation at all.

Discuss Grading

Discuss is a feature within Nualang in which teachers can assign an interpersonal speaking exercise to groups of students, using an online meeting tool to discuss the assigned topic in the language of their course. The students can record all or part of their discussion and submit it for grading. The teacher can then view the recording and grade the contributions of the students. When grading, the teacher has an option to ask for an initial AI grading, which can be updated or discarded. The grading uses established rubrics appropriate to the course to provide grades under various categories.

Whether or not AI is used in the grading, the recordings are never exported.

Services Used

The size of data sets and cost of compute resources makes it prohibitive for organizations other than mega-corporations to train the large models. In implementing generative AI functionality in our products, we do use third party generative AI tools. Our software architecture does allow us to switch between providers and use multiple providers if required. For all providers that we use, we always ensure that our service agreements provide that any data we send to the provider is not retained and is not used to train their models. Notwithstanding these service agreements, we do not, in any case, send any private or personal data to any third party providers, ensuring that our data and our customer's data remains secure on our own systems.

In providing services to our customers that are based on AI, we use Retrieval Augmented Generation (RAG), where the capabilities of Large Language Model (LLM) of the third party provider is augmented by our own data and customer's data, which is only stored on our own servers and not sent to the LLM provider.

Summary

To summarize, we have been adding AI capabilities to our products in carefully considered ways to help our staff and our customers to get more done more quickly, while always retaining editorial control.

We place great emphasis on data privacy and safeguarding PII and don't export such data from our own servers. In particular, string substitution of personal identifications occur in all queries sent to our AI code and the reverse substitution is applied to responses before they are returned to users. Psuedo random placeholder identifications, such as "student adxprt" are used in this process.

We do not train models or allow models to be trained using our customer's data.