As a byproduct of FY26 funding, the following is a breakdown of the AI services that will initially be delivered to campus. They are organized by service tier which describes our means to scale their availability to the campus (e.g. equitable – broadly available, targeted – to enable use cases with the greatest impact, and specialized – to support application development and promote research).
- Equitable – Microsoft Copilot Chat with Enterprise Data Protections (EDP)
- Targeted – Open AI ChatGPT for Edu
- Specialized – AI gateway service residing within a cloud environment
Copilot Chat with EDP is an AI service provided by Microsoft which is now available to all students, faculty, and staff at UWM. As cited by our use cases, this application can support a variety of personal and organizational productivity needs. Because it is bundled with existing licensing, the University does not incur additional costs from Microsoft to offer this service. Copilot Chat can be accessed through the web (https://copilot.cloud.microsoft) utilizing PantherID credentials. Although this service is available to all populations, usage caps limit how much it can be continuously used. Because Microsoft has branded the term Copilot for all AIenabled offerings in their portfolio, there are additional Copilot applications now available which are not part of Copilot Chat licensing. For example, Copilot for M365 requires additional licensing not in scope for Copilot Chat.
ChatGPT for Edu is an optimized instance of ChatGPT by OpenAI for higher education. Based upon our use cases, we know this service is highly sought out for a variety of purposes. This license is typically used to support coursework and projects. Due to the cost of this licensing, UWM will maintain a finite pool to be allocated each semester as needed based upon need and availability. This is a common practice in higher education due to the unsustainable cost at scale of this service. This initial pool will consist of 300 licenses and is expected to be available for use by fall 2025.
Establishing an AI gateway service within a cloud environment is a common approach for how institutions can establish an environment capable of developing custom GPTs, Retrieval Augmented Generation (RAG) chatbots, AI agents, and other AI-enabled applications at scale using a variety of AI models. These needs reflect several use cases cited by the Taskforce. This model is far more cost-effective and capable at scale than conventional AI subscription services such as Copilot Chat and ChatGPT Plus if you have the talent required to build, operate, and support these platforms. There are several different types of AI gateways that have been developed as commercial or open-source platforms over the last couple of years. Because we have some limited expertise with Microsoft Azure, this will be the preferred cloud environment to support these services. This will become clear as we finalize the selection of this platform guided by the requirements to deliver these ROIs, collaborations with other institutions to discover best fit, and guidance by vendor partners with expertise in these technologies. Funding also provides professional services to aid in this implementation and cross-training of talent to operate and support this new environment.
It is important to note that a continuous flow of commercial and open-source AI products have emerged in this market. As a result, there is a rapidly evolving and growing set of new capabilities and elevated performance characteristics. Because of this, licensing and cost models have also been evolving at a rapid pace. As a result, we anticipate there may be enhancements or alternatives to consider that will likely alter this initial portfolio over time as we seek to adopt this technology in safe and responsible ways.