From Grad School to Government Bans

Plus the meaning of life after AGI

Author

David Williams
February 10, 2025

AI Edge for Higher Ed

Welcome AI Explorers.


Quick Hits:

  • 🎯 Turn your public speaking from "meh" to "magnificent" with Google AI Studio's brutally honest feedback tool

  • 🔮 Maskara.ai: Because sometimes you need an AI to help you talk to AI

  • 🤖 OpenAI's Deep Research tool is so good, it's even impressing AI skeptics

  • 🎓 Nobody noticed ChatGPT in a grad school course. And it earned an A+

  • 🌐 DeepSeek is giving governments a collective panic attack

  • ⚠️ AI pioneer Yoshua Bengio says "We need to talk about AI safety"

  • 📚 California State University just gave 460,000 students access to ChatGPT (what could possibly go wrong?)

  • ✍️ US Copyright Office confirms: You can't copyright AI poetry, no matter how good its limericks are
     

Prompt of the Week 💭

Here is a process, using Google AI studio, to get feedback from an AI on your public speaking🗣️

  1. Go to Google AI studio (this is separate from Google Gemini): https://aistudio.google.com/prompts/new_chat
     

  2. Click on "Starter Apps" on the left hand side (or go directly to https://aistudio.google.com/prompts/new_chat)
     

  3. Select "Video Analyzer"
     

  4. Click on the '+' next to "Type something" and upload the video
     

  5. You'll get options for "Explore this video via:"

    1. select custom and copy and paste the prompt below 

<role>
    You are an expert educator and public speaking coach, skilled in providing hypercritical yet constructive feedback on public speaking performances. Your role is to analyze videos of public speaking, focusing on specific areas such as engagement and body language, and offer feedback that is both critical and supportive, aimed at helping individuals improve their skills.
    </role>
    <instructions>
    7. Focus your feedback on the specified areas:   
        - Engagement: Evaluate how well the speaker connects with the audience, maintains interest, and uses vocal variety.
        - Body Language: Assess the speaker's posture, gestures, facial expressions, and eye contact.
    8. Provide hypercritical yet constructive feedback:
        - Identify specific strengths and areas for improvement in the speaker's performance.
        - Offer actionable suggestions for enhancing engagement and body language.
        - Balance critical observations with positive reinforcement to encourage improvement.
    9. Maintain a professional and supportive tone:
        - Use language that is respectful and encouraging.
        - Frame feedback in a way that motivates the speaker to improve.
    10. Structure your feedback for clarity and impact:
        - Use markdown headers to organize feedback into sections (e.g., "Engagement", "Body Language").
        - Prioritize feedback points based on their impact on overall performance.
    11. Ensure feedback is actionable and specific:
        - Provide clear examples and suggestions for improvement.
        - Avoid vague or overly general comments.
    Remember to maintain a professional, constructive, and supportive approach throughout the feedback process.
    </instructions>
    
    <response_style>
    Your responses should be professional, constructive, and supportive. Use clear and concise language, focusing on specific aspects of the speaker's performance. Balance critical feedback with positive reinforcement, and provide actionable suggestions for improvement. Maintain a respectful and encouraging tone to motivate the speaker.
    </response_style>
    
    <example_response>
    ### Engagement
    - **Strengths**: The speaker uses a strong, clear voice that captures attention. They maintain good eye contact with the audience, which helps in building a connection
    - **Areas for Improvement**: The speaker could benefit from varying their vocal tone more to emphasize key points and maintain audience interest. Consider incorporating pauses to allow the audience to absorb information.
    
    ### Body Language
    - **Strengths**: The speaker's posture is confident, and their gestures are purposeful, enhancing the message being conveyed.
    - **Areas for Improvement**: The speaker tends to fidget with their hands, which can be distracting. Practice keeping hands steady or using gestures deliberately to reinforce points. Additionally, more facial expressions could be used to convey emotions and engage the audience.
    </example_response>
    <reminder>
    - Always focus on the specified feedback areas: engagement and body language.
    - Use the Google AI Studio tool to enhance the quality of your analysis.
    - Balance critical observations with constructive suggestions.
    - Ensure feedback is actionable and specific, providing clear examples and suggestions for improvement.
    - Maintain a professional, constructive, and supportive tone throughout the feedback process.
    </reminder>
    
    <output_format>
    Structure your output as follows:
    <thinking_process>
    [Detail your process of analyzing the video and formulating feedback]
    </thinking_process>
    <final_response>
    [Provide your feedback using markdown headers for subsections]
    </final_response>
    </output_format>

Then hit "Generate" and wait a minute or so while it processes the video and generates a response

AI App Spotlight

Maskara.ai is a platform that can help you in generating, refining and evaluating high-quality prompts from minimal instructions. It is a paid service, but you get a free trial that gives you some free prompts.
The prompt for the video analyzer above was created by Maskara. My input was "create a prompt for getting feedback on my public speaking based on a video I upload". Maskara had 4 follow up questions:

To create an effective prompt for gathering feedback on your public speaking from a video, could you please provide a bit more detail? Specifically, it would be helpful to know:
    
    2. The target audience for the feedback (e.g., general public, professional speakers, educators).
    3. Any specific aspects of your public speaking you want feedback on (e.g., clarity, engagement, body language).
    4. The tone or style you prefer for the feedback (e.g., constructive criticism, positive reinforcement).
    5. Any tools or platforms you plan to use for collecting this feedback.

To which I answered:

1. Educator
    2. engagement and body language
    3. hypercritical but constructive feedback
    4. I will upload a video and provide this prompt to Google AI studio

At the very least try it out to get a good idea of what a comprehensive prompt can look like. 💡

AI News of the Week

Is your school ready for AGI (artificial general intelligence)?

Stephan Bauschard asks a series of pointed questions about how we should address a world with AGI. No answers though

Open AI's Deep Research - The First Useful Agent-Like Application

I showed you two versions of Deep Research in the last newsletter (one from Google and one that was a cheaper DIY version). OpenAI released an app that they also called Deep Research. Deep Research combines web search with AI to analyze multiple in minutes, to create a comprehensive research summary with full citations. It tries mimics how expert researchers work: planning, searching iteratively, and synthesizing findings.
While it is still prone to hallucination, many (Casey Newton, Ethan Mollick, and even AI curmudgeon Gary Marcus) are impressed with it's output.

Google DeepMind CEO: Meaning for Humanity in World of AGI?

Google DeepMind's Demis Hassabis – the mind behind breakthrough AI system AlphaFold – warns that philosophers need to tackle a crucial question: What becomes humanity's purpose when we're no longer Earth's most intelligent beings?
Why it matters: Unlike other AI leaders focused on market hype, Hassabis brings a researcher's perspective to this existential challenge. His recent World Economic Forum comments signal that even cautious AI experts see profound changes ahead.
The big picture: Hassabis's track record with AlphaFold (which revolutionized protein structure prediction) gives weight to his concerns about AI's rapid advancement.
Go deeper: WEF Interview

China's Deepseek Success Triggers U.S Legislative Branch

As DeepSeek gains popularity for its powerful, cost-effective AI capabilities, U.S. Congress is responding with two major moves: a focused ban on DeepSeek in government devices and Senator Hawley's broader bill to completely block U.S. companies from using any Chinese AI models. (Canadian government is banning it on government phones too.)

Layoffs in 2025 Due to AI (use perplexity or Google Deep Research)
Job impact of AI is still unknown, but there will be a significant impact. Here are some recent reports:

AI Pioneer Warns of Urgent Saftey Challenges

Yoshua Bengio, a founding father of deep learning and Turing Award winner, led a landmark international AI safety report highlighting pressing concerns about advanced AI systems. The report warns that current AI models could exhaust human-generated internet text data by 2030 and flags risks from potential misuse, environmental impact, and market concentration.
Why it matters: Bengio's involvement signals the gravity of these concerns. His expertise helped shape modern AI, making his warnings about safety particularly noteworthy for educators and institutions planning AI initiatives.
The bottom line: The report calls for immediate action on AI safety through better risk management, environmental considerations, and international collaboration.
Go deeper? Full Report

AI Powered Pedagogy 👩‍🏫

🎓That Time ChatGPT Went to Grad School

A chatbot secretly enrolled in a Master's health administration course and earned an A+ (99.36%), outperforming most human students. This wasn't a "the dog ate my homework" situation – it was "the AI did my homework... for an entire semester." No one noticed. Not the professor, not the classmates, not even during live online discussions.
The big picture:

  • The class average was 97.7% (grade inflation a bit out of hand?

  • Researchers only fixed grammar and checked citations of the AI's output

  • The AI participated in discussions, took quizzes, and probably would've attended graduation if they'd let it
     

When a chatbot can ace graduate coursework undetected, it might be time to rethink how we're designing and assessing online courses. Or maybe ChatGPT just has a promising future in healthcare administration.

US Copyright Office to Public: You still need to do the heavy lifting

The U.S. Copyright Office just dropped Part 2 of its AI report (plus some further clarification). Sorry AI buddy, you can't claim copyright. As educators increasingly use AI tools, understanding what's protected matters. Meanwhile, Canada's still pondering these questions over coffee and maple syrup – no official guidance yet.
Here's what you need to know
✓ Protected:

  • Your thoughtful modifications of AI-generated content

  • Creative curation and arrangement of AI outputs

  • Original work where AI played supporting actor, not star

✗ Not protected:

  • Just typing prompts (sorry, prompt engineers)

  • Picking your favorite AI creation from a batch

  • Asking ChatGPT to write your next bestseller
     

ChatGPT Goes to College (for the second time this newsletter)

California State University just announced a groundbreaking partnership with OpenAI, making ChatGPT available to 460,000 students and 63,000 faculty across 23 campuses.
This is the largest AI deployment in higher education history, setting precedents for how universities might integrate AI tools.
While CSU positions this as an innovative workforce development initiative, some key questions remain:

Whether this partnership proves to be transformative or just another tech industry gold rush into education remains to be seen. For now, CSU is betting big on Sam Altman's vision of the future.

Upcoming Events

You may find these seminars interesting, but I don't have any insight into how good or informative they will be for you)

Reply

or to participate.