AI Resources

How Can I Use AI? 

CIO Mairéad Martin and Technology Services (TS) announced the availability of Microsoft Copilot with Data Protection last week. TS also pointed to helpful information about how to get started with Copilot which seems best suited for staff and administrative work. 

In anticipation the more sophisticated AI needs of instructors and researchers in the Grainger College of Engineering, Engineering IT (EngrIT) has secured access to enterprise class AI cloud services. If you would like to make use of these resources, please reach out to our IT Architect, Michael Chan. He has investigated generative AI platforms and shares some topics of consideration below based upon his research and experiences.

Is it ethically or morally right to use AI?

We obviously can not answer that for you, the Grainger College of Engineering, or for broader audiences considering the possibilities of misuse, misinformation, bias, etc. We do believe the choice should be personal, wisely considered, and realistically made, especially in an academic environment. It is important to understand the risks, and to take the warning that all various AI platforms present seriously. We found the article Ethical and Responsible Use of Generative AI in Scholarly Research provided a balanced assessment of the benefits and practical advice on how to address risks.

Purchasing and Procurement

University purchasing and procurement rules apply to using cloud platforms and software. OpenAI, the most popular platform, wants to store a credit card so it can bill for usage. This process is not permitted by University purchasing policy or state law. We suspect some folks have used p-cards to pay for OpenAI and other cloud services, and we would encourage those folks to work with their business office team to comply with the University rules and regulations around purchasing. Personal purchases of cloud services are not reimbursable. 

Even free cloud services require agreement with some contract, which technically only a handful of people at the University of Illinois are permitted to do on behalf of the University. Copilot, previously Bing Chat Enterprise, circumvents these purchase and contractual missteps because it is already included in services we contract from Microsoft.

Data Privacy Biggest Concern

Data privacy is probably the biggest concern. Unless clearly stated otherwise, one must assume that anything put in the cloud is going to be seen by other people and companies are going to use that data. "Enterprise" versions of AI generally build data privacy into their contracts. ChatGPT, even the paid version, and Google Bard state they will use the data users provide in the use of their AI clients. 

Standing up independent deployment and instance of AI in cloud-based services like Azure, AWS, and Google Cloud Platform is usually a safer option. These companies generally do not mine that data, and it is generally encrypted at rest and in flight. Each service agreement should still be read carefully to understand the data privacy aspect of any deployment. There is still always a chance other humans could see your data, but it is small.

Copilot, previously Bing Chat Enterprise, explicitly states that it does not keep or use the data you share with it.

Further Reading

University of Illinois, Office of the Provost, Generative AI Center of Expertise

Technology Services, Privacy and Cybersecurity published Privacy Considerations for Generative AI that help provide general guidance on use of AI at the University of Illinois Urbana-Champaign.

EDUCAUSE has also provided suggestions for Integrating Generative AI into Higher Education.