Understand the Implications of Generative AI in Education

Bans aren't the answer

Analyst Perspective

Guide the appropriate use of generative AI in teaching, learning, and working.

Generative AI has emerged as a prominent technology, garnering substantial attention in a very short time. The advanced functionality of generative AI for searching and creating content has raised concerns about its implications in education. Info-Tech Research Group has undertaken an examination of the impact of generative AI on education, offering insights into the future implications for educators, students, and IT professionals.

The use of generative AI raises concerns about its effects on education and student learning, with cheating being a primary concern. Nonetheless, the technology has the potential to transform the classroom just as calculators, the internet, and digital assets changed the way we taught and learned in the past. Educators may have no choice but to embrace generative AI and leverage its strengths to enhance learning and development.

Mark Maby
Research Director for Education, Industry Practice
Info-Tech Research Group

Understand the implications of generative AI in education

Generative AI

• ChatGPT reached 100 million users in just two months after launching.
• ChatGPT is generative AI, and like other technologies before, it will have substantial ramifications on education.
• Understand the implications of generative AI in the short, middle and long term.

Identify high-value processes

• Identify your capabilities that support key business processes based on the whether they are conversational, repeatable, and valuable.
• Build a heatmap, highlighting areas where a generative AI solution could be effective.

Info-Tech Insight
IT will need to provide guidance on the proper response to generative AI, which will change how we teach, learn, and work in education.

OpenAI and ChatGPT: An overview

“I think it's incumbent on those of you in IT … to spread the word broadly, not in a semester, but now. We need to let our clients know what this is, what the future is likely to be, and how they may use it without jeopardizing the quality or rigor of their teaching.”
Ray Schroeder on ChatGPT,
UPCEA Senior Fellow and UIS Professor Emeritus (Educause, January 30, 2023).

Short term tactics: ChatGPT and Cheating

What have been the responses to ChatGPT among faculty and leadership at your institution?

Blocking sites on the institution’s network
Prominent institutions have blocked ChatGPT on their network. This has mostly been at the K-12 level, including New York City PS in the US and Queensland and NSW in Australia (Education Week). Sciences Po in France is an example at the university level (Reuters, January 27, 2023).
Limited Effectiveness: 95% of students have mobile phones to access ChatGPT (Pew Research Center).
Risks: Prevents creative use of the tool by innovative teachers.

Plagiarism-detection tools
Tools can assess authenticity of a text as human-written or AI-generated. Turnitin is a prominent example but there are other tools, including one offered by OpenAI. The company is even an attempting to statistically “watermark” its output (TechCrunch).
Limited Effectiveness: These tools provide a confidence metric, and AI will likely improve its “naturalness”.
Risks: These tools come with their own privacy issues. Also, students will become suspicious of their teachers who act suspicious toward them.

Blocking sites at the level of the device
Higher education testing centers and labs can be used as dedicated centers for written assessment where access to ChatGPT is blocked.
For instructors grappling with flagrant use this may be an option.
More effective ways of assessment likely exist, such as oral discussion.

Mid-term strategy: Leveraging AI in education

Does it seem to you that AI writing tools will demand greater student/instructor interaction? Can institutions afford this?

Generative AI will cause a sea change to literacy education

Problematic approach - continue to question the authenticity of students’ writing

• Can we continue to use writing as a measure of achievement?
• Continuation of term papers may come under scrutiny.
• Second-language students are a test case.
• Good writing will come under suspicion.
• Poor writing will be penalized.
• Using AI to grade student essays.
• Just as instructors question the students’ writing, students will question their instructors’ feedback.

Positive approach - increase ownership for the positions expressed in the writing

• Turning a generated essay into the students’ own version.
• Use the tool to teach editing skills.
• Learn to shape and personalize the output.
• Flipped classroom so that seminal pieces of work are done in class.
• Focus more on in-class writing, oral presentations, and multimedia assignments.
• AI literacy so that students think critically.
• For example, show how societal biases are embedded in the technology.

Mid-term strategy: Leveraging AI in education

Do you know of academics using the tool already in your institution?

AI can alleviate the administrative burden on faculty

A faculty member works about 61 hours a week in higher education (Research Gate) and a teacher, 54 hours in K-12 (Education Week).

• About 30% – 50% of that time is spent in administrative duties which aren’t student-facing or related to research pursuits.
• This is where generative AI can be especially supportive.
• Academic use cases of generative AI in higher education (Mushtaq Bilal).
• Outline for a research/seminar paper (see example on following slides)
• Brainstorm research questions
• Draft a personal statement for a scholarship/graduate school
• Interview preparation
• Presentation preparation
• Teaching use cases of generative AI in K-12 education (HerftEducator).
• Lesson planning (see example on following slides)
• Essay exemplars
• Peer and self-review assessments
• Essay feedback
• Rubric creation

Risks to automating these parts of teaching

• Factual errors
• Grading students with AI may break trust with students

AI + Human advantage

• AI is recommended for first draft outlining
• The output is noted for lacking empathy and insight, which the teacher and faculty can bring to the text through revision.

Mid-term strategy: An outline for a graduate seminar created by ChatGPT

Prompt:
Topic: Canada-US relations during last 20 years
Context: Graduate seminar at School of Advanced International Studies
Requirement: Outline for a seminar paper with research questions
Language: Academic
Tone: Formal

1. Introduction

1. Background on Canada-US relations
2. Importance of the topic
3. Research questions

2. Overview of Canada-US relations during the last 20 years

1. Major events and developments
2. Key players and their roles
3. The impact of these developments on Canada-US relations

3. Economic relations

1. Trade agreements between Canada and the US
2. Economic dependence and interdependence
3. Tariff disputes and their impact

4. Security and defense cooperation

1. Military and intelligence cooperation
2. Border security and immigration policies
3. Counterterrorism efforts

5. Energy and environmental policies

1. Energy trade and environmental regulations
2. Keystone XL pipeline controversy
3. Climate change policies and cooperation

6. Cultural and social relations

1. Public opinion and attitudes toward each other
2. Cross-border mobility and cultural exchange
3. Diversity and multiculturalism

7. Challenges and prospects for Canada-US relations

1. Current challenges and obstacles
2. Opportunities for cooperation and improvement
3. The role of leadership and diplomacy

8. Conclusion

1. Summary of key findings
2. Implications and recommendations for future research
3. Policy implications for Canada-US relations

Research Questions:

1. How have Canada-US relations evolved in the last two decades, and what are the main drivers of this evolution?
2. What are the economic implications of the trade agreements and tariff disputes between Canada and the US in recent years?
3. How have security and defense cooperation between Canada and the US developed since 2001?
4. What are the main energy and environmental policy issues that have affected Canada-US relations, and how have they been addressed?
5. What role have cultural and social factors played in shaping Canada-US relations, and how have they evolved over the last 20 years?
6. What are the main challenges facing Canada-US relations in the present day, and what opportunities exist for cooperation and improvement?

Mid-term strategy: A K-12 lesson plan created by ChatGPT

Prompt:
Create a lesson plan.
Subject: Science
Topic: The Water Cycle
Grade: 4
Duration: 1 class period (45-60 minutes per class period)

Objectives:

• Students will be able to explain the different stages of the water cycle.
• Students will be able to identify the different forms of precipitation.

Materials:

• Pictures and diagrams of the water cycle
• Whiteboard or chalkboard
• Markers or chalk
• Worksheets

Procedure:

Introduction (5 minutes)

Begin the lesson by asking students what they know about the water cycle. Write their responses on the board.

Introduce the objectives of the lesson.

Stages of the Water Cycle (20 minutes)

Show pictures and diagrams of the water cycle.

Explain the different stages of the water cycle (evaporation, condensation, precipitation, and collection).

Have students label and color a diagram of the water cycle.

Forms of Precipitation (20 minutes)

Introduce the different forms of precipitation (rain, snow, sleet, hail).

Show pictures and diagrams of each form of precipitation.

Have students identify the different forms of precipitation and describe how they are formed.

Assessment (10 minutes)

Have students complete a worksheet to check their understanding of the lesson.

Assessment:

Worksheets to check understanding of the lesson.

Conclusion:

By the end of the lesson, students should have a better understanding of the stages of the water cycle and the different forms of precipitation. The lesson plan includes a variety of activities and assessments to engage students and ensure that they have a comprehensive understanding of the water cycle.

Mid-term strategy: Leveraging AI in education

Administrative opportunities

Chatbots for use both internally and externally

15% of institutions are using chatbots ¹⁵ for admissions support, but these are largely based on scripted responses (Educause, February 16, 2023).

Chatbots will be much more effective with generative AI, not only with both prospective students but also current students, alumni, faculty, and the larger community.

Greatest technical challenges to AI adoption as identified by CIOs in education:

• Ineffective data management and integration 72%
• Insufficient technical expertise 71%
• Financial concerns 67%
• Immature data governance 66%

There is a mistrust of leveraging AI for efficiency (IPSOS).

What is the cultural perception of using generative AI for administrative duties?

Long-term governance: Policy on AI in Education

What have been the responses to ChatGPT among faculty and leadership at your institution?

Areas of concern

• Governance for data and privacy
• Ensuring equal access and opportunities
• Curriculum innovation to address the implications of AI
• Financial support for the effective implementation of AI (AI and Education)

Policy considerations

• How can AI be leveraged to enhance education?
• How can we ensure the ethical, inclusive and equitable use of AI in education?
• How can education prepare humans to live and work with AI?

Approaches

• Independent approach, where there is a specific AI policy
• Integrated approach, where AI is addressed with other technologies
• Thematic approach, where AI is addressed with a related topic such as GDPR

Few countries are ready

• Many countries are drafting legislation to address AI
• Europe is furthest ahead with the AI Act (Center for Data Innovation)
• The legislation will have implications for organizations using AI
• Most countries have not yet acknowledged the effect AI will have on learning

Long-term governance: Policy on AI in Education

How would AI legislation affect your department’s adoption or support of AI tools?

Up-coming AI Act in the European Union

The AI Act will regulate AI tools and services

• Subordinate to existing industry regulation (such as for medical devices)
• Non-compliance can result in fines or withdrawal from market

“Limited risk” AI tools / services

• Systems must comply with GDPR when collecting PII
• Disclosure of deep-fakes, except for artistic or satirical purposes

“High risk” AI tools / services

• Generative AI may be placed in a “high-risk” category
• AI “providers” must demonstrate conformity within the new framework
• Risk management
• Testing for risks
• Data governance controls
• Technical documentation
• Sufficient transparency for user to interpret the output
• The AI has human oversight
• Organizations using AI for ‘customers’ – including institutions of education
• Need an AI risk management system for documentation and mitigation
• Apply transparency measures when AI interacts with its customers

“Unacceptable risk” tools / services are banned due to their threat to society

Source: Center for Data Innovation.

Other countries are less further along than the EU:
U.S.: Algorithmic Accountability Act of 2022 (Charles Morgan et al.)
Canada: Bill C-27 Digital Charter Implementation Act 2022 (Schwartz Reisman Institute).

What’s the next step for IT?

Refine your use case for generative AI.

1. Review the business capability map for high-value processes
1. Identify your capabilities that support key business processes.
2. Build a heatmap, highlighting areas where a generative AI solution could be effective using a framework of whether a process is repeatable, generative, and valuable.
2. Conduct a basic cost-benefit analysis for the tech
1. Of the most appealing processes/capabilities, select one or a small number that are most suitable for generative AI intervention.
2. Outline a high-level business case — How would generative AI reduce manual work? What would its output look like? What would the expected benefit be? What are the drawbacks? How much would it cost?
3. Explore the vendor landscape
1. Identify the table stakes — What does your AI solution need to do? What solutions does it need to integrate with?
2. Review options available on the market. Many tools exist but are not "products." ChatGPT is available as a research preview while other tools, like Jasper, Rytr, and the various ITSM chatbots, can all be purchased for enterprise use.
3. Compare your requirements to your options, select, and begin piloting.

Leverage Info-Tech’s business capability maps to identify high-value processes

Business capability map defined…

In business architecture, the primary view of an organization is known as a business capability map.

A business capability defines what a business does to enable value creation with the following characteristics:

• Represent stable business functions and processes.
• Are unique and independent of each other.
• Will typically have a defined business outcome.

A business capability map provides details that help the business architecture practitioner direct attention to a specific area of the business for further assessment.

Here is an example of the Business Capability Map for Higher Education. Editable examples of this map and the K-12 map are included below.

Business Capability Map for Higher Education

Applying the framework to higher ed

Map the capabilities from the previous page to the Venn diagram based on the following criteria.

Generative or conversational:

• Is this the sort of work a generative AI can do?
• Does it involve conversation or the creation of new material?

Valuable:

• Would generative AI meaningfully reduce manual effort in completing a task?
• Can generative AI be used without extensive permissions, contextual expertise, or institutional risk?

Repeatable:

• Is the task repeatable and would it occur frequently enough to train a model and recoup initial investment?

Higher education capabilities highlighted for benefit from generative AI

Review use cases: Chatbot for admissions in universities

Description

Colleges and universities may use chatbots to enhance the admissions experience by providing quick access to information, be available 24/7, and handle a large volume of inquiries simultaneously. Those using generative AI or other types of machine learning can provide personalized responses based on student profiles. Chatbots can reduce the workload of admissions officers and provide a cost-effective solution for managing the admissions process. Using chatbots can create a better experience for prospective students, increase efficiency, and save the institution money.

Specific use and description of output: Generative AI could be used as a chatbot to answer the questions of prospective students about the admissions process, academic programs, campus life, and other important aspects of life at the university or college.

Expected benefit: The chatbot can also be available 24/7, handle a large volume of inquiries, reduce the workload of admissions officers, and provide a cost-effective solution for managing the admissions process.

Possible drawbacks: The technology is new and untested, sometimes provides the wrong information, and would likely need human review before going in front of a prospective client.

Cost: Pricing varies by vendor, but generally users are billed by the word, by a monthly fee, or a combination of the two. However, the cost of implementing a chatbot increases with its complexity, the level of personalization required, and the amount of programming and maintenance needed. In the long run, using a chatbot for admissions can be a cost-effective solution compared to hiring additional admissions staff.

Business capability map for K-12 education

Applying the framework to K-12

Map the capabilities from the previous page to the Venn diagram based on the following criteria.

Generative or conversational:

• Is this the sort of work a generative AI can do?
• Does it involve conversation or the creation of new material?

Valuable:

• Would generative AI meaningfully reduce manual effort in completing a task?
• Can generative AI be used without extensive permissions, contextual expertise, or institutional risk?

Repeatable:

• Is the task repeatable and would it occur frequently enough to train a model and recoup initial investment?

K-12 capabilities highlighted for benefit from generative AI

Review use cases: Virtual assistant for K-12 parent communication

Description

By leveraging generative AI as a virtual assistant for parent communication, school districts can improve parent engagement, reduce administrative workload, and provide a better overall experience for parents. The district would have to train the virtual assistant on a large corpus of data, including transcripts of past parent communication, school policies and procedures, and student data. It is important to ensure that the virtual assistant is transparent, secure, and respects the privacy of both parents and students.

Specific use and description of output: Generative AI could be leveraged as a virtual assistant to improve communication with parents. The virtual assistant can answer parents' questions about the school district's policies, procedures, and resources, and provide updates on student progress.

Expected benefit: The virtual assistant can provide quick and accurate responses to their queries, while reducing the workload of school administrators and freeing up time for other important tasks.

Possible drawbacks: There may be concerns about the accuracy and reliability of the virtual assistant's responses, particularly in cases where the inquiry is complex or involves sensitive information.

Cost: Pricing varies by vendor, but generally users are billed by the word, by a monthly fee, or a combination of the two. However, the cost of implementing a virtual assistant increases with its complexity, the level of personalization required, and the amount of programming and maintenance needed. In the long run, using a virtual assistant for parent communication can be a cost-effective solution for districts with strained resources.

Explore the vendor landscape and conduct a RASF

Review the market space once you understand your desired future state and broad requirements.

SoftwareReviews is an excellent resource for this type of research.

Pictured here is the Data Quadrant for chatbot solutions, but your requirements may align with other software categories.

Leverage the Rapid Application Selection Framework methodology to begin your selection analysis.

Rapid Application Selection Framework (RASF)

Awareness > Education & Discovery > Evaluation > Selection > Negotiation & Configuration

Related Info-Tech Research

The Five CIO Priorities for 2023
Engage cross-functional leadership to seize opportunity while protecting the organization from volatility. Included among the top five CIO priorities is the directive to prepare your data pipeline to train AI.

AI Governance
Leverage Info-Tech’s framework to effectively manage and govern your AI deployment.

Build a Chatbot Proof of Concept
Investigate whether a chatbot will support your students.

Bibliography

Arai, Maggie. “Five Things to Know about Bill C-27.” Schwartz Reisman Institute. Accessed February 17, 2023.

Atske, Sara. “Teens, Social Media and Technology 2022.” Pew Research Center: Internet, Science & Tech (blog), August 10, 2022.

Browne, Ryan. “Microsoft Reportedly Plans to Invest $10 Billion in Creator of Buzzy A.I. Tool ChatGPT.” CNBC, January 10, 2023.

Carmichael, Matt. “We Used ChatGPT to Help Us Poll Americans about AI. Here’s What We Found.” IPSOS, January 2023.

Metz, Cade. “Inside OpenAI, Elon Musk’s Wild Plan to Set Artificial Intelligence Free | WIRED.” Web Archive. April 27, 2016.

Romero, Alberto. “GPT-3 — A Complete Overview.” Towards Data Science, May 25, 2021.

OpenAI. “ChatGPT: Optimizing Language Models for Dialogue.” OpenAI, November 30, 2022.

Hu, Krystal. “ChatGPT Sets Record for Fastest-Growing User Base - Analyst Note.” Reuters, February 2, 2023, sec. Technology.

Schroeder, Ray. “Member QuickTalk | GPT: The Generative AI Revolution.” Webinar presented by Educause, Online, January 30, 2023.

Klein, Alyson. “New York City Blocks ChatGPT at Schools. Should Other Districts Follow? Banning the AI-Powered Writing Technology Does Students a Disservice, Some Experts Say.” Education Week, January 18, 2023. Gale Academic OneFile.

Miao, Fengchun, Wayne Holmes, Ronghuai Huang, and Hui Zhang. AI and Education: A Guidance for Policymakers. UNESCO Publishing, 2021.

Mueller, Benjamin. “The Artificial Intelligence Act: A Quick Explainer.” Center for Data Innovation (blog), May 4, 2021.

Morgan, Charles S., Francis Langlois, and Jerry Lan. “U.S. House and Senate Reintroduce the Algorithmic Accountability Act Intended to Regulate AI.” McCarthy Tétrault, April 8, 2022.

Reuters. “Top French University Bans Use of ChatGPT to Prevent Plagiarism.” Reuters, January 27, 2023, sec. Technology.

Kurtz, Holly. “A Profession in Crisis: Findings From a National Teacher Survey.” Education Week, April 14, 2022, sec. Teaching Profession.

Mushtaq Bilal, PhD [@MushtaqBilalPhD]. “ChatGPT Is Everywhere and Everyone Is Using It.” Twitter, February 3, 2023.

Herft, Andrew. “A Teacher’s Prompt Guide to ChatGPT.” HerftEducator, January 17, 2023.

“EDUCAUSE QuickPoll Results: Artificial Intelligence Use in Higher Education.” Educause, Accessed February 16, 2023.

Wiggers, Kyle. “OpenAI’s Attempts to Watermark AI Text Hit Limits.” TechCrunch , December 10, 2022.

Ziker, John, et al. “Time Allocation Workload Knowledge Study, Phase 1 Report,” Research Gate, October 2, 2013.