TODAY: NVIDIA will make its AI supercomputers in the US in what some are calling a win for the Trump tariffs; OpenAI releases its next major update, but there's a catch that might annoy web users; and we tackle the critical thinking dilemma that AI has created through (what else?) more prompt tips.

APRIL 15, 2025

• Steal This Prompt: 🧠 AI Hurts Critical Thinking UNLESS You Prompt Like This
• Top Story: NVIDIA to Make AI Supercomputers in U.S. for the First Time
• THE HEADLINES
• TOOL OF THE DAY: ​GPT-4.1 Is Now Available in API Form

STP: 🧠 AI Hurts Critical Thinking UNLESS You Prompt Like This

A 2025 study by Microsoft Research and Carnegie Mellon University warns that growing confidence in generative AI can erode critical thinking. After surveying 319 knowledge workers and analyzing 936 real-world GenAI use cases, researchers found a troubling pattern: users who relied heavily on AI were less likely to think critically about their work.

Three key areas were impacted:

• Information gathering: AI reduces search effort, but users often skip verifying the results.
• Problem-solving: Instead of independent thinking, users tweak AI responses.
• Task execution: Users oversee rather than engage, shifting from active doers to passive editors.

In short, AI isn't just doing our tasks—it may be dulling our mental edge. That should concern anyone using AI to write, ideate, or solve problems.

But it doesn’t have to be this way.

Prompt engineering—the same skill blamed for this mental atrophy—can actually overcome it.

If used with intention, prompts can sharpen how you question, verify, compare, and reflect. They can teach you to think harder, not just work faster.

​STEAL THESE 5 PROMPT TIPS TO USE AI WITHOUT LOSING YOURSELF IN THE PROCESS​

📰 NVIDIA to Make AI Supercomputers in U.S. for the First Time

In a major shift toward domestic production, NVIDIA announced it will manufacture its AI supercomputers entirely in the U.S. for the first time. The company is collaborating with partners like TSMC, Foxconn, Wistron, Amkor, and SPIL to build and test Blackwell AI chips in Arizona and assemble full AI supercomputers in Texas.

Production of NVIDIA’s new Blackwell chips has already begun at TSMC’s Phoenix facilities. Meanwhile, Foxconn and Wistron are preparing factories in Houston and Dallas, with mass production slated to scale up within 12–15 months.

NVIDIA says this U.S.-based expansion could produce up to $500 billion in AI infrastructure over the next four years, creating hundreds of thousands of jobs and strengthening national supply chain resilience. These supercomputers will power “AI factories”—a new class of data centers dedicated to AI processing.

CEO Jensen Huang emphasized the significance: “The engines of the world’s AI infrastructure are being built in the United States for the first time.” NVIDIA will also employ its own AI and robotics platforms—including Omniverse and Isaac GR00T—to design, simulate, and automate operations at these advanced facilities.

​CLICK HERE FOR MORE ON WHY​

THE HEADLINES

• ​Amazon CEO Andy Jassy shares how AI will reinvent ‘virtually every customer experience we know’​​
• ​How Google AI is helping decode dolphin communication​
• ​3 New Ways AI Agents Can Help You Do Even More​
• ​Meta to Use Public Posts, AI Interactions to Train Models in EU​
• ​AI-driven UV technique offers rapid contamination check for cell cultures​​
• ​CDS Assistant Professor Merges AI and Behavioral Science​
• ​IBM: Supply Chain Execs See Competitive Edge in Agentic Generative AI​
• ​Palantir Stock Pops After NATO Acquires AI-Enabled Warfighting System​
• ​Tassel Is a New AI Catching On with Colleges: Its Specialty Is Reading Student Names at Graduation​
• ​Sam Altman: OpenAI Has Reached Roughly 800 Million Users​
• ​Donatos Pizza Selects RevMo AI for Voice Ordering​
• ​AI company Hugging Face buys humanoid robot company Pollen Robotics​​

TOOL SPOTLIGHT: ​GPT-4.1 Is Now Available in API Form

​

On Monday, April 14, OpenAI introduced GPT-4.1, its latest AI model, along with two streamlined versions: GPT-4.1 Mini and GPT-4.1 Nano.These models offer enhanced capabilities in coding, instruction comprehension, and processing extensive information, all at reduced costs.

🔍 Key Features

• Advanced Coding Proficiency: GPT-4.1 demonstrates a 21% improvement over GPT-4o and a 27% enhancement over GPT-4.5 in coding tasks, making it a valuable tool for software development and debugging.​
• Enhanced Instruction Understanding: The model excels at interpreting and following complex instructions, reducing the need for repetitive inputs and streamlining task execution.​
• Expanded Context Processing: With the ability to handle up to 1 million tokens, GPT-4.1 can analyze and summarize large documents or datasets more effectively than its predecessors.​
• Cost and Speed Efficiency: GPT-4.1 operates 40% faster and is 80% more cost-effective per query compared to GPT-4o, offering significant savings and efficiency gains.​

💼 Practical Applications

• Software Development: Assists in writing, reviewing, and optimizing code, enhancing productivity for developers.​
• Customer Support: Powers AI agents capable of resolving customer inquiries with minimal human intervention, improving response times and service quality.​
• Data Analysis: Facilitates the extraction of insights from extensive reports or datasets, aiding in informed decision-making.​
• Educational Tools: Enables the creation of personalized learning applications, such as language flashcards or interactive tutorials.

🧠 Model Variants

• GPT-4.1 Mini: Offers a balance between performance and cost, suitable for applications requiring efficient processing without compromising on capabilities.​
• GPT-4.1 Nano: Designed for tasks demanding rapid responses and minimal resource usage, ideal for real-time applications and mobile platforms.​

OpenAI's GPT-4.1 models are accessible through their API, providing businesses with versatile tools to enhance various operations. No word yet on when Free, Plus, and Pro web users will have access.

For more detailed information and access, visit OpenAI's official announcement​, or watch the full 23-minute video linked above.

​Read All mAIn Street Back Issues Here​

Your input fuels the content. Let me know how I did today!