Sponsored by
Hello readers,
Welcome to the AI For All newsletter! Today, we’re talking about how college professors are leveraging AI to force their students to defend their arguments, the use of generative AI in enterprise and IoT applications, and more!
AI In Action: Forcing you to defend yourself
The arrival of AI in the classroom has created an obvious problem: students who once had to develop their own arguments have been showing up to class with someone else’s. But a growing group of professors has decided the answer to AI-assisted laziness isn't to ban AI — it's to build better AI, The Washington Post reports. At Columbia Business School, one professor at Columbia Business designed a debate-prep tool called Caisey that, instead of handing students an answer, argues back. Students come in with a position, Caisey pushes them on it, and the conversation only ends when they've actually defended their thinking.
The approach has deep pedagogical roots. The Oxbridge tutorial model — where a student is challenged by an expert until their reasoning is genuinely stress-tested — has long been considered one of the most effective forms of education ever devised. But it’s proven difficult to scale. Faculty-designed AI tools are now making a credible run at that problem, with Caisey already deployed at 16 institutions and Georgia Tech running its own version for engineering students.
Where ChatGPT will summarize a case study on demand, these tools are built around specific course materials, calibrated to specific learning goals, and deliberately engineered to slow students down rather than speed them up. Students reported learning more deeply, making fewer repeated mistakes, and — perhaps most surprisingly — actually enjoying the process.
The most valuable AI tools may not be the ones that make work faster. They may be the ones designed to preserve what makes us human: the capacity to actually think.
🔥 Rapid Fire
Analysis: The Subprime AI Crisis Is Here
Consumers and businesses are paying subsidized rates for AI
The subscription model does not work for LLMs
Charging the actual costs would guarantee massive churn
AI companies impose unpopular rate limits instead
Reduced model quality often accompanies rate limits
Pricing is fixed despite the product’s value fluctuating
Every AI startup is unprofitable — including OpenAI and Anthropic
There was only ~$60B in total AI revenue in 2025
Almost $1T in CapEx alone has gone into AI so far
Venture capital is facing a historic liquidity crisis
Venture capital cannot sustain AI indefinitely
Analysis: How Much of the AI Bubble is Real?
The fake Disney-OpenAI deal is just one of many
NVIDIA, SK Hynix, Samsung, AMD, Broadcom — all fake
Increases in token consumption do not indicate rising AI adoption
Recent models are simply more token-intensive
Only 5 GW of data center capacity is “under construction” globally
That’s 5 GW out of the 200+ GW announced
The reality is that most AI data centers will never be built
AI data center company Fermi’s shares plunge 13% on $486M net loss
Fermi’s 11 GW development halted construction in February 2026
About half of US data centers due in 2026 will face delays or cancellations
Oracle stock plummets on severe cash crunch and shrinking margins
Oracle’s credit risk measure hits record high on AI debt fears
Microsoft closes worst quarter since 2008 on AI concerns
Addendum: the slowing growth of SaaS is not because of AI
OpenAI shares have fallen out of favor on the secondary market
Investors trying to sell $600M of shares couldn’t find buyers
OpenAI says prompt injection may never be solved for browser agents
An indeterminate amount of judges are using AI despite unreliability
Google introduces Gemma 4 open models with Apache 2.0 license
Ship Docs Your Team Is Actually Proud Of
Mintlify helps you create fast, beautiful docs that developers actually enjoy using. Write in markdown, sync with your repo, and deploy in minutes. Built-in components handle search, navigation, API references, and interactive examples out of the box, so you can focus on clear content instead of custom infrastructure.
Automatic versioning, analytics, and AI powered search make it easy to scale as your product grows. Your docs stay accurate automatically with AI-powered workflows with every pull request.
Whether you're a dev, technical writer, part of devrel, and beyond, Mintlify fits into the way you already work and helps your documentation keep pace with your product.
📖 What We’re Reading
By training models on historical IoT data, organizations can generate predictive scenarios, simulate edge cases, and create synthetic datasets for testing that would be impossible or dangerous to produce in real-world conditions. A smart irrigation system, for example, can be tested against AI-generated data simulating sensor malfunctions or power spikes — without exposing actual hardware to failure conditions.
Beyond testing, generative AI is being applied to anomaly detection, predictive maintenance, and natural language interfaces for IoT dashboards. The convergence is creating a new class of intelligent connected systems.
