- There Is No Spoon
- Posts
- Chronicles from the Vibe Economy
Chronicles from the Vibe Economy
Transistor cafes, journalism compromises, and AI that costs a lot more than your coffee habit

AI Edge for Higher Ed
"Is ChatGPT-4.5 worth the money?" That's just one question we tackle in this week's newsletter, alongside practical teaching prompts you can use tomorrow, fresh research on who's winning (and losing) in the AI workplace revolution, and my slightly obsessive adventures in "vibe coding."
🤖 Teaching Prompts - Socratic dialogues and critical thinking prompts for more engaging learning
🚀 ChatGPT-4.5 - The good, bad, and expensive: worth your prompt budget?
📰 NYT & AI - How journalism trying to balance AI efficiency with human authenticity
💼 Workplace Impact - AI boosts productivity, especially for younger, educated workers
🎓 Education Framework - Ten considerations for AI integration in universities
🖥️ Vibe Coding - Build software just by describing it (I made a Transistor Café Website in 5 minutes!)
📅 Upcoming Events - Webinars on AI writing, learning assessment, and bias
Prompt of the Week 💭
Some Teaching prompts via Robin Delta on x.com () https://x.com/heyrobinai/status/1886355123544441252)
Socratic Enlightenment
Guide me through a Socratic dialogue on [concept]. Ask insightful, probing questions that help me explore the deeper layers of this idea and gradually refine my understanding step-by-step.
Teach-back Mastery
I will attempt to teach you the concept of [topic]. Please correct any inaccuracies in my explanation and ask clarifying questions that deepen my understanding of the material.
Critical Thinking
Challenge me with a series of thought-provoking questions on [subject]. After each question, offer feedback on my answer—highlighting both strengths and areas for improvement—and suggest actionable ways to enhance my critical thinking on this topic.
Master Any Topic
Act as an expert in [subject]. Briefly explain the key concepts, including a real-world example for each. Then, provide a concise, step-by-step 30-day plan to master the topic, outlining daily or weekly milestones, suggested resources, and practical exercises
AI App Spotlight
ChatGPT-4.5 is here (for some users) and it promises better conversations, broader knowledge, and fewer mistakes - but it costs WAY more to run it. ChatGPT-Pro users ($200US/month) currently have access to it and ChatGPT-Plus users ($20US/month) will get access this week. Free users may have access at some point, but that's uncertain right now.
I currently have access through the API, but due to its high cost, I am rapidly burning through my monthly allotment of prompts.
User reviews are mixed, even Sam Altman (CEO of OpenAI) has tried to temper expectations saying it "feels like talking to a thoughtful person" but it "won't crush benchmarks".
What Users Like:
More intuitive, engaging conversations.
Enhanced creativity in writing and problem-solving.
Improved reliability with fewer inaccuracies.
What Users Dislike:
High cost
Limited complex reasoning and logic capabilities.
It's kinda slow
Incremental improvements rather than groundbreaking innovation.
Still trails competitors like Claude 3.7 Sonnet in creativity and humor
GPT-4.5 is a solid upgrade for smoother, more creative conversations—but its steep price and modest improvements mean it's not a game-changer (yet). If you're looking for groundbreaking innovation, you might want to wait. For now, keep an eye on your prompting budget—I know I am!
AI News of the Week
Can Journalism Survive the AI Revolution. The New York Times puts it to the test with an updated AI use policy
The New York Times is now cautiously embracing AI allowing its use for edits, headlines, and interview question suggestions (but decidedly not for article writing). Journalism is built on trust, human judgement and authenticity, so can journalism maintain its soul in an AI-driven future? Or will efficiency quietly erode the trust that makes news valuable?
(full article).
Interestingly, I couldn't find NYTimes statement on how it is allowing AI use, just articles from lots of other sources. I did find this statement from May 2024 though.
Generative AI at Work: Who Wins, Who Loses, and What's Next?
Nearly one-third of U.S. workers (4000+ surveyed) now use generative AI tools on the job. These workers are typically younger, higher educated, and higher income. Their generative AI use was found to triple productivity in some tasks. As AI adoption accelerates at an uneven rate, how will it impact job security, skills needed and wages. This study probes this questions, but unsatisfyingly has no solid answers.
(full paper)
More Generative AI at Work: Focusing on Customer Support
This recent study of customer support workers shows how generative AI assistance significantly increases worker productivity—especially benefiting less skilled and less experienced employees. Closely following AI recommendations leads to the greatest productivity gains, enabling human agents to communicate more effectively and reducing customer requests to speak with supervisors. However, the study also highlights potential risks, including over-reliance on AI and decreased human ability to tackle novel problems.
(full paper)
AI Powered Pedagogy 👩🏫
Generative AI in Irish Higher Education: Ten Critical Considerations
Irish universities (along with universities around the world) are increasingly adopting generative AI tools raising questions around academic integrity and educational quality. This paper outlines ten key considerations and a frameowrk for addressing the responsible integration of AI into learning and teaching. (Spoiler: AI lteracy, Allowable AI, Academic Integrity, Critical AI, Second-degree Plagiarism, Privacy, Equitable access, Sustainability, AI sovereignty, and Enhancing Teaching and Learning.
(full paper)
Citing AI Isn't Enough: Introducing the Artificial Intelligence Disclosure (AID) Framework
As generative AI tools like ChatGPT become integral to research and education, traditional citation methods fall short in capturing their complex and varied roles. This article introduces the Artificial Intelligence Disclosure (AID) Framework to transparently document AI's involvement across the research and writing process. The AID Framework can help maintain academic integrity, clarify AI's contributions, and address critical gaps in current citation practices.
(full article)
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). For many of these webinars, a recording and resources will be sent to all registrants - so even if you can't make it live, they can be worth registering for.
AI for Writers Summit. Virtual Event March 6, 2025. Targeted more for marketing, but even if you never want to use AI for writing, you can get a sense of what heavy users are doing with generative AI tools (e.g., things that your students might be doing)
Debunking the AI Snake Oil. March 27, 206 12:00-1:00. Free but sponsored
Checking Progress and Evaluating Learning with AI. March 12. 12-1:30. From UBC, so it requires a UBC login - if you've ever been a student there, you will have one.
AI's Hidden Fears: Navigating Inherent Bias & Hallucinations. Mar 11. 3:00-4:00. From University of Delaware
More from University of Delaware: https://ats.udel.edu/ai-workshops/
The mAIn Event
The mAIn Event: Vibe Coding
"Vibe coding," is an approach to software development where you use AI to build software by simply describing your ideas to it.
How Vibe Coding Works
You describe what you want—"Make me a plant-lover's Instagram"—and the AI produces working code without you writing a single line. When something's off (that green is definitely more "hospital hallway" than "lush garden"), you simply point it out and the AI adjusts. Need a new feature? Just say "add plant identification" and it appears, no stuggling with syntax required on your part. It's software development where your biggest technical challenge is finding the right words to explain your vision—not debugging mysterious errors at midnight.
The Double-Edged Sword of Vibe Coding
Vibe coding makes software development remarkably accessible—describe what you want and watch it materialize without writing a single line of code. Build in hours instead of weeks, test ideas instantly, and focus on solving problems rather than wrestling with syntax. The convenience is undeniable, but so are the risks: unpredictable code quality, potential security blind spots, and indecipherable results that even the creator can't understand.
Perhaps most concerning is the use of vibe coding by those learning how to code — we're perhaps creating a generation of developers who can build anything but understand nothing, capable of summoning digital castles but helpless if the AI ever goes offline. It's like learning to "cook" exclusively with meal delivery kits—convenient until you actually need to understand the ingredients.
Getting Started
Choose an AI tools: The most popular option is Cursor with Claude 3.7 Sonnet, but another good option is to use the artifact feature built into Claude 3.7 directly. Available even in the free version.
Pick a small project: Start simple, like a calculator or basic web app.
Craft clear prompts: Be specific and concise in your descriptions.
Iterate gradually: Break projects into small steps and refine as you go.
Test thoroughly: Regularly check functionality and user interactions.
Me, Vibing
My favourite tool to use for Vibe Coding is "Create with Code" on SimTheory. Here are some of the things I've created (each took me less than 10 minutes):
Website for a Bipolar Junction Transistor Cafe (it's like a cat cafe, but for bipolar junction transistors)1: https://simulationtheory.ai/24e019d6-1f69-44fe-bb8e-bcf7e08dd2a5
Book Smart Game - Catch the books, get smarter: https://simulationtheory.ai/26719e12-d7cb-4731-9ff5-1daf740d53f2
Snake Chase Game - Avoid the kids, collect the apples: https://simulationtheory.ai/a698524c-0f02-4250-b765-d5f551b5f758
The website addresses look weird because SimTheory uses a random number generator to make sure each application has a unique URL.
The bottom line
Vibe coding makes software development more accessible and efficient, although the easier it becomes to build software without understanding the code, the more vulnerable we become to its limitations.
1: I teach a course on electronic devices and a big component (pun intended) of the course is Bipolar Junction Transistors
Reply