AI Goes to College

In this wide-ranging discussion, Craig Van Slyke and Robert E. Crossler explore recent AI developments and tackle the fundamental challenges facing higher education in an AI-enhanced world. They begin by examining GPT Tasks, highlighting practical applications like automated news summaries and scheduled tasks, while sharing personal experiments that demonstrate the importance of playful exploration with new AI tools.

The conversation then turns to Gemini's new fact-checking features, with important cautions about source verification and the need to balance convenience with critical evaluation of AI-generated content.

The hosts have an engaging discussion about the challenge of "transactional education" - where learning has become a points-for-grades exchange - and explore alternative approaches like mastery-based learning and European assessment models. They discuss concrete strategies for moving beyond traditional grading schemes, including reducing assignment volume and focusing on process over outcomes.

The episode concludes with an announcement of an upcoming repository for AI-enhanced teaching activities and a call for educators across disciplines to share their innovative approaches.

Key Takeaways:

• GPT Tasks enables automated, scheduled AI interactions - from news summaries to daily content delivery
• Gemini's new fact-checking feature provides source verification but requires careful evaluation of source credibility
• Deepseek, the new AI model, is worth checking out but be aware of privacy concerns
• The challenge of "transactional education" requires rethinking traditional assessment methods
• Practical alternatives to points-based grading include focusing on mastery and reducing assignment volume
• Faculty across disciplines are invited to contribute to a new repository of AI-enhanced teaching activities

Outline

GPT Tasks and Functionalities

• AI agents conducting various tasks.
• Use case: AI travel planner and email agent.
• Development of GPT tasks to run scheduled prompts.
• Example uses:
• Receiving AI and higher ed news updates.
• Daily dad joke feature.

Exploration of New AI Tools

• Importance of experimenting with AI tools for learning.
• Application play and psychological basis for learning technology.
• Encouragement to try new tools without overcomplicating the process.

Comparison of Search Tools

• Comparing ChatGPT with Google alerts.
• Tailoring information relevance and accuracy.
• Importance of validating AI-generated information.

Privacy and Availability of AI Tools

• Availability limited to certain user levels and regions.
• Variability in tool features across different platforms.

DeepSEEK: A New AI Model

• Introduction and capability of DeepSEEK.
• Cost efficiency and openness as an open-source model.
• Privacy concerns related to data sharing and Chinese government access.
• Impact on NVIDIA stock and benchmarks comparison.

Open Source and Computational Needs

• Role of open source in future AI model development.
• Computational requirements and challenges with local running versions.

Privacy and Intellectual Property Concerns

• Distinction between privacy and intellectual property.
• Concerns about research data compliance and institutional rules.

Writing with AI Tools

• Differentiating writing and editing.
• Using AI to enhance rather than replace human creativity.
• Increase in writing quality in published work.

Transactional Education Model

• Challenges posed by transactional nature of education.
• Importance of mastery and process focus in learning.
• Discussions on final exams vs. continuous assessment.

Proposed Repository for Active Learning Activities

• Building a shared resource for educators.
• Encouragement for community contribution and collaboration.
• Long-term vision for enhancing educational engagement with AI.

Conclusion and Call for Interaction

• Inviting listeners to contribute ideas and share practices.
• Encouragement for community-driven improvement in AI-integrated education.

Time Stamps:

[00:00] Introduction and GPT Tasks discussion

[15:45] Gemini's new features and source verification

[25:20] Writing process and AI tools

[35:10] Transactional education challenges

[45:00] Announcement of teaching activity repository

Links

• ChatGPT: https://chat.openai.com/
• Gemini: https://gemini.google.com/
• DeepSeek: https://deepseek.ai/
• Anthropic: https://www.anthropic.com/
• Claude: https://www.anthropic.com/claude
• Llama: https://ai.facebook.com/blog/large-language-model-llama-meta-ai/ (This links to a blog post about Llama, as there isn't a dedicated website for the model itself.)

Mentioned in this episode:

AI Goes to College Newsletter