I’ll share which AI tools I’m exploring, some experiments I’m conducting, and insightful information about what I’m observing in the world every week.

🔧 Three Tools I’m Testing

📊 Napkin.AI - Creating charts, graphs, visuals for text that is always a challenge for the graphically challenged. Napkin makes it easier to take text and build fun graphics.

💬 Google AI Studio - A comprehensive set of tools and APIs, but I use it to transcribe speech to text, for free. I’ve been looking for a simple free way to transcribe random audio files - this will be my go to for a while.

🧑‍💻 Claude 3.7 - Anthropic’s latest model is even better at coding. This translates to improvements in Cursor and playing around with Claude Code. I’m deep in building out my “vibe coding” workflow and this makes it better.

🧪 AI Experiment of The Week

I have a fair bit of old video from my youth (and not so youth) including DVDs of my college football highlights. Now, the quality of these is what you’d expect from recordings from the early 2000s. However, I’d love to try to upscale these at some point. Well, this week, I played around with a tool for using AI to improve the fidelity of lower-resolution videos. The tool is Starlight from Topaz Labs, it’s currently in experimental phase, but I was impressed. You can enhance three videos a week for free.

I decided to test it using an old video I took of the Grand Canyon on a trip I took about a dozen or so years ago. Using the browser site, you have to upload MP4 formatted video, so I just used a simple converter to go from .mov to .mp4. Then you upload the video and click “Render” and it goes to work. About 20 minutes later, you get an enhanced version. Below is an image of the video's two versions (original on left, enhanced on right). Even from this image you can easily see the improved details on the right.

I was impressed with the improvements. You can watch the videos here. A couple things to mention about the current state of the service: you can only upload three videos per week; each can be no more than 300 frames. The videos you upload on the free plan can be used for their training. There is a paid early access where you can do up to five minutes of video per render and it costs 90 credits, equating to roughly $10 to $20 depending on how you purchase the credits. However, once I pull my DVD highlights, I’ll try a clip to see the quality, and if it is solid, I will convert the rest at cost.

📰 Article of The Week

Tech Execs Blast the EU About AI Regulations - a dive into how Google, Meta, and other executives came down hard on European AI regulations.

AI regulations are vital for many reasons. First, we need to ensure we’re providing guidelines for safely allowing rapid innovation for our AI companies. The US is currently (and usually always) a bit more laissez-faire than Europe. The current administration wants the US to be the world leader in AI, and the companies here would love to be a part of that. However, we also need to make sure the best AI tech stays in the country, which means additional regulations on export and limiting access to technology (like those latest NVIDIA chips). This means there’s a delicate balance around consumer safety, innovation, and export control.

The tech giants argue to Europe that their AI Act, which was created very early on in the AI boom, is driving down innovation. Additionally, they are worried that the regulations will keep them from providing AI products in the market, which will hit their bottom line. So, what’s the answer? Personally, I don’t know, but it’s probably less restrictive than the current European AI Act.

🌎 Where the World is Going

The rapid advance of AI promises transformative benefits across healthcare, education, and productivity. Yet as we eagerly input our financial records, medical histories, and personal conversations into chatbots and AI agents, we're creating unprecedented vulnerabilities in our digital privacy landscape.

It's striking to consider how limited public information was before the internet era. Unless featured in media, most people maintained relative anonymity—the yellow pages and door-knocking salespeople represented the extent of "data collection." Today, with minimal effort, anyone can uncover phone numbers, family details, and personal habits. AI accelerates this exposure exponentially by processing, remembering, and potentially leaking everything we share.

Where does this privacy erosion end? While I don't have definitive answers, I increasingly find myself strategizing about maintaining boundaries between my public and private digital lives. In the coming years, expect a booming industry dedicated to extracting personal information from AI models—similar to today's services that remove Social Security numbers from data broker sites. The personal data revolution 2.0 isn't coming—it's already here, and we must prepare accordingly.

👨‍💻 About Me

🔗 Extra Links to Check Out

• Chatbot Arena Leaderboard 🎯

• Raytheon’s AI Radar Software Improvements ✈️

• ElevenLabs Scribe - Speech-to-Text 🗣️