Recommendations
Recommendations
Generative Artificial Intelligence Task Force Report
To aid TMU in achieving its broader goals and support the proposed framework for AI adaptation, the task force makes recommendations in the following five areas:
Establish university-wide AI principles and the development of guidelines for learning & teaching, SRC, and administrative uses.
We recommend the adoption of the proposed principles above. These should then inform the development or updating of guidelines for learning and teaching, SRC, and administrative uses of AI. Guidelines in each area should be established by the appropriate committees and offices. Where guidelines already exist, they should be updated to ensure alignment with the proposed principles.
Update relevant Senate and administrative policies to ensure alignment with institutional principles on generative AI.
The “Policies & Guidelines" section above provides a list of current senate and administrative policies that may govern or be implicated by AI use at the university. These policies should be reviewed to update language and ensure they are aligned with the suggested institutional principles for AI adaptation at TMU. In particular, policies and guidelines should have strong and clear language concerning disclosure and transparency.
Create an advisory committee on artificial intelligence.
We recommend establishing an advisory committee to support the Provost and the Vice President, Administration and Operations with implementing recommendations from this task force, monitoring progress, sharing results, and further engaging the community. The composition of this committee should include leaders with primary responsibilities identified in this report including the Office of the Vice President, Research and Innovation, the Office of the Vice-Provost Students, the Office of the Vice-Provost Academic, the Chief Information Officer, and Dean of Libraries. This could be established as a subcommittee of the Advisory Committee on Academic Computing (ACAC). There may also be a need for a dedicated role to coordinate the activities of this committee and provide direction on issues in AI in higher education.
Create an AI web portal.
The university should build upon the existing AI website using it as a portal to collate institutionally supported AI tools, policies, guidelines, training, and other resources to support work across all functional areas.
Review license agreements for restrictions on AI use.
The Libraries must continue to provide guidance to the TMU community on the use of copyrighted and licensed material with AI tools, while at the same time assessing license agreements to minimize unnecessary restrictions on researchers’ abilities to text and data mining (TDM). Furthermore, the Libraries should actively participate in national advocacy efforts to prevent vendor contracts from overriding established fair dealing rights.
Monitor sector practice and legal status of copyright and AI.
The university should ensure that fair dealing rights are appropriately described, exercised and upheld by continuing to monitor the evolving legal landscape related to copyright and AI. Guidance on such matters should continue to be updated and evolve with the changing policy environment and emerging case law. The Libraries, General Counsel, and Central Computing Services should continue to collaborate on updated guidelines.
Establish secure AI environments for experimentation and innovation.
Across learning and teaching, SRC, and operations, there is a need for locally hosted AI tools to experiment and innovate. We recommend that TMU investigate the feasibility of establishing locally hosted AI sandbox tools, using multiple models, including open source models (see example of Vanderbilt University’s Amplify platform (external link) and models provided by the TMU Libraries Collaboratory). This could allow for a range of custom AI applications and innovations and could have benefits for the development of TMU’s research and entrepreneurship ecosystem. In addition to making these tools available, we recommend the establishment of secure, isolated, and locally hosted virtual environments (e.g. similar to U of T’s “AI kitchen”) to facilitate experimentation with these technologies in strict compliance with privacy regulations, safeguarding data leaks and intellectual property.
Collaborate on shared platforms and computing infrastructure.
The COU AI Task Force Report (external link) (p. 21) provides excellent guidance on the potential advantages of Ontario universities collaborating on the development of shared multi-model AI platforms and high-capacity computing infrastructure. TMU should lead and participate in any such consortium arrangements. There are also opportunities for both provincial and federal investments in sovereign AI compute capacity for which a collaborative approach may prove advantageous.
Enable expanded access to advanced institutional AI tools.
Our Google Workspace AI tools have premium tiers and additional advanced features. We recommend that these be made available at cost to users (similar to expanded Google Drive storage options). This expanded access could also include Gemini API access for advanced research purposes, application development, and administrative uses. This could also include support for AI coding assistants for software development. Researchers also require extensive latitude for AI experimentation and applications, which should allow for the use of a range of AI tools following robust privacy and data security assessments. In order to encourage experimentation TMU’s financial policies should take a flexible approach to the reimbursement of AI tools beyond Gemini.
Exercise prudence and caution in the adoption of new enterprise AI software and services.
Enterprise AI software and services are nascent technologies. Hundreds of new educational technology companies have developed products for universities from AI tutors to chatbots to student information systems and more. TMU should exercise extraordinary caution when evaluating the adoption of new AI technologies as these software and services are still in early development and there is a high degree of risk of vendor lock-in. Many of these new products are of unclear value and quality. The long-term outlook for such software and services is uncertain and any adoption of new tools should undergo rigorous evaluation, erring on the side of caution. There is no urgency to making new investments in enterprise AI software and services while the market is still in an uncertain growth phase. TMU would be best positioned to monitor and observe developments across the sector and among other large enterprises to further assess where strategic investments may best be made.
Adhere to clear procurement protocols for AI technologies.
The university should establish the following criteria for adopting any AI tools:
- Prohibition on Training: Absolute contractual prohibition against the vendor utilizing TMU data, records, inputs, or prompts to train public or proprietary models.
- Legislative Compliance: Explicit alignment with provincial and federal privacy legislations, (i.e. FIPPA), employment law (Employment Standards Act (ESA), human rights law, alongside professional licensing obligations.
- Audit Logging and Trails: Comprehensive capabilities for tracking, logging, and auditing all AI interactions to ensure absolute accountability.
- Core System Compatibility: Any tools adopted outside or alongside core platforms must feature verified, safe, and cost-effective API compatibility with TMU's underlying technology stack and legacy core Enterprise Resource Planning (ERP) systems, specifically Oracle EBS.
- Accessibility compliance: ensuring AODA requirements and TMU accessibility standards are met.
- Vendor submission of third party reports (i.e. PIA, TRA, SOC 2) and demos of AI service / technology where feasible and/or appropriate.
- Contract / vendor management provisions addressing notice of changes, disclosure, compliance with industry standards, privacy / security, submission of updated third party reports, vendor accountability.
Offer comprehensive AI fluency training opportunities.
There is a need for a range of AI fluency training opportunities for faculty/contract lecturers, staff, and students to advance AI readiness and ensure equity of opportunity. Training should be designed for the needs and roles of different users at the university, supporting learning and teaching, SRC, administration, and student career preparation services. These opportunities should be offered across the university, led by the appropriate service providers including CELT, TMU Libraries, OVPRI, HR, CCS, The Chang School, and Career Services.
Create internal AI knowledge base.
We recommend the creation of a central repository for AI documentation, guidance memos, tool updates, and industry-specific advancements to serve as a resource for university leaders, faculty/contract lecturers, and staff. This will support users across the university with access and information about institutional AI tools and future developments of these technologies.
Expand and enrich AI communities of practice.
In 2023, the AIO established an AI community of practice (external link) with a focus on learning and teaching. This operates largely as an email list for sharing resources. We recommended expanding AI communities of practice across the university based on the three strategic focus areas of the framework identified above (Adaptation and Innovation in Learning and Teaching, Acceleration of Scholarly Research and Creative Activity, and Safe and Secure Operational Efficiencies). During our consultations, we heard from many members of our community a desire for more opportunities to meet, discuss, and share ideas, concerns, and questions about AI in all university operations. The conversations we had with colleagues during the consultations were excellent and demonstrated a need for continued community engagement and collaboration on these matters.
Continue and expand support for digital learning, teaching development, and curriculum development programming in CELT.
CELT offers a range of programming to support faculty/contract lecturers with adapting learning and teaching to generative AI. This includes programs like the Digitally Enhanced Learning Partnership Grants and Students as Partners in Generative Artificial Intelligence program. It also includes long-standing programming that aids faculty/contract lecturers with course design and assessment planning. These services are essential to facilitating the redesign of the assessment of learning in programs across the university. Furthermore, services from the Curriculum Quality Assurance unit in CELT support programs with reviewing and updating program-level learning outcomes to make discipline-appropriate curriculum changes to adapt AI competencies and skills in academic programs.
Expand support for assessment security and academic integrity.
The commercial release of ChatGPT and other AI chatbot services has had an immediate effect on academic integrity in the learning environment in higher education across the sector. Traditional forms of academic misconduct including plagiarism and contract cheating have seen significant decline while academic misconduct involving the use of AI chatbots to misrepresent one’s authentic knowledge and skills has grown significantly, challenging the ability of instructors to authentically assess student learning. Faculty/contract lecturers are deploying a variety of strategies to adapt methods of assessment to this new reality and we recommend that resources continue to be devoted to services that support such adaptations. There is a heightened need for secure assessment environments, which has led to a growth in in-person assessments and exams. TMU will need to expand capacity to support in-person assessments both during the formal exam periods and during the Fall and Winter semesters for midterm exams. We also recommend that the university continue with the advice of the Academic Integrity Office (AIO) to avoid the use of AI detection software as it remains unreliable. Instead, we advise continued support for the educational initiatives and outreach activities of the AIO and Student Life and Learning Support (SLLS).
Support pedagogical research on artificial intelligence in higher education.
While there have been immediate effects of AI on academic integrity, further impacts of these new technologies on the learning environment remain uncertain. The university can play a leading role in expanding knowledge and understanding of the effects of AI on learning and teaching in higher education through expanded pedagogical research. TMU’s teaching fellows program is an excellent model for supporting pedagogical research, including research that investigates the effects of AI in the learning environment in universities. We recommend that this support continue and that the university appoint fellows whose work focuses particularly on generative artificial intelligence.
Explore the integration of AI in student success services.
Integrating AI into student-facing services such as academic accommodations, well-being supports, and registrarial functions presents an opportunity to enhance service delivery and improve the overall student experience. Strategic use of AI tools can help reduce wait times for routine inquiries, streamline processes related to admissions and transfer credits, and offer timely support in areas like academic advising. The university should also position AI as a tool that enhances human support but does not replace it. Use AI to streamline administrative processes, improve student engagement, and reduce staff workload, while retaining and safeguarding human judgment in sensitive areas such as mental health or advising.
Integrate accessibility and inclusive design in AI adaptation.
AI tools must meet accessibility standards and be compatible with assistive technologies. Policies and accommodations should formally recognize AI’s role in supporting students with accommodations without penalizing its use under academic integrity frameworks. If AI-supported accommodations are made available, they should be explicitly referenced in university policies with clear guidance for faculty to distinguish between legitimate accessibility use from misconduct.
Create an AI Action Lab as a neutral cross-disciplinary hub connecting faculty and students experimenting with AI in scholarly research and creative activity.
The university could establish an AI Action Lab where identified innovators in scholarship and professional practices are supported in developing an AI project contextualized within their own scholarly or education disciplinary culture. Participants would be provided with access to tools, expertise, and a structured community to engage in cross-disciplinary exchange, knowledge sharing, and collaboration in critical experimentation of AI. The university could consider appointing an AI SRC specialist role to coordinate the Lab. This Lab could be modeled on proven initiatives like Columbia University’s aiX Faculty Fellowship Program (external link) , and University of Virginia’s AI Literacy and Action Lab (external link) . The University may consider building upon the strengths of existing innovation spaces such as the TMU Libraries’ Collaboratory.
Build or adopt AI tools for grant proposal development.
Building on projects like Automated Grant Feedback at Western University (external link) , TMU should consider piloting a similar tool available for researchers to provide them with feedback on draft grant proposals. Such tools have the potential to expand service capacity to support faculty research and improve grant application success rates.
Expand and update research ethics training on AI.
AI technologies present some unique challenges for research ethics. The university should expand and update research ethics training on AI. This could include training for TMU researchers as well as members of TMU’s research ethic boards.