By Ryan Gravette, IDLA Technology Director
I often get the question, “How do you keep up with AI?” The answer is hours of research and testing, most of it outside the workday. Below, I outline how I am doing that currently:
Keeping up with AI is a substantial commitment, one that extends beyond the typical work week. It requires evenings, weekends, and even drive time (Podcasts) to stay informed. When technology evolves at such a rapid pace, it’s our dedication to understanding and growth that compels us to invest time outside the traditional workday. Here’s how I do it, and more importantly, how I find fulfillment in the process.
Research
There are a few blogs from special people. They are special because they are leaders in their field and willing to share what they learn. It takes energy to learn, and even more energy to share. I appreciate them and am taking Fridays to share back in the same way.
Keeping up with AI in Education – Technical Complexity medium
Ethan Mollick is from the Wharton School of Business, where he performs research and teaches. His LinkedIn posts, blog, and book are all world-class in terms of understanding trends in AI and the impact on education. His daily updates keep you up to date with what is going on in AI, and his weekly blog posts go deep into trends and topics impacting the field. Here is a quote: “People assume that AI homogenizes creative writing, producing much less diverse work than groups of humans. This paper finds this isn’t true: given stories to complete, GPT-4o writes as diversely as humans (stylistic, lexical, & semantic) when prompted with context & randomness.”
The EdTech Insiders podcast and newsletter are spot-on regarding what is happening in EdTech. Led by Alex Sarlin, the group consists of leaders from across the industry, and while they focus on edTech in general, recently, AI has dominated much of the conversation. The team does a great job of breaking complex topics down and discussing how it impacts education. Here is a recent quote from an article to wet your curiosity: “Right now, this competition is playing out most visibly in what we’re calling the ‘Battle of the Study Modes.” OpenAI, Anthropic, and Google have all launched dedicated learning experiences within their flagship products, representing a fundamental shift in how AI companies think about education.
This is a big deal for a few reasons:”
Keeping up with Programming with AI – Technical Complexity High:
Before the Commit – Danny Gershman & Dustin Hilgaertner outline how they use AI to accelerate programming. They try new tools and discuss the value proposition in good detail. They also have a cybersecurity segment, “Secure or Sus,” where they talk about current security events surrounding AI. I ask all our developers to listen to this podcast.
Zak El Fassi – Zach blogs frequently and thinks deeply about AI and the future of AI. He wrote the wonderful Talk and Comment tool for K12 and continues to fund it at his cost. His insights are a foresight into what is happening now and what the future could look like. It is both philosophical and tactical. His words are often technical poetry. Here is a snippet: “Here’s what kills me: we’re building AI agents, distributed systems, and elaborate protocols for machine consciousness on top of email infrastructure designed in 1971. We’re constructing digital cathedrals on foundations made of rusty pipes full of lead.” And I anticipate he is working to fix that. His recent posts on MCP are dead on in my mind.
Keeping up with AI in General –
The Dwarkish podcast and blog by Dwarkish Patel push the edges of AI by using history to inform and question AI’s path. Wonder how the Japanese wars before WWII impact how we think about AI. I listened to the podcast, and there is great stuff there. Dwarkish’s unique approach of using historical events to analyze and question the trajectory of AI provides a fresh perspective and valuable insights. Here is a quote: “Sometimes people say that even if all AI progress totally stopped, the systems of today would still be far more economically transformative than the internet. I disagree. I think the LLMs of today are magical. However, the reason that the Fortune 500 aren’t using them to transform their workflows isn’t because the management is too stodgy. Rather, I think it’s genuinely hard to get normal humanlike labor out of LLMs. And this has to do with some fundamental capabilities these models lack.”
Intelligent machines, part of the TWiT network and hosted by Leo Laporte, Jeff Jarvis, and Paris Martineau, provide both interviews with AI leaders and news in AI. The hosts do a great job aggregating the news on AI and discussing the impacts and relevance. If you are interested in how AI scored against humans in a creative writing competition and how it has changed since GPT4 to GPT5, tune in. They break it down. If you want that article about the writing competition they discuss in Episode 833, check the link below. In traditional clickbait style: I was shocked and saddened by the result.
Testing
I have been using Claude Code for around three months and have been impressed by the results. I built an app in around 140 hours that would have taken me 3 years to finish. I use it every week for new tasks and activities. The most recent was to create a POC middleware between Qualtrics and Agiliz Buzz. The week before, it was to create a report of users and settings across the Agilix Buzz LMS using the Agilix Buzz API. Every time I throw something at it, it surprises me. I encourage everyone to test and play with the tool, it is and feels like the future.
I recently started using MCP services to see what was possible. Today, for example, I used Claude to create an Asana project to track a recent contract we signed with SHI. I took the contract and asked Claude to use the MCP connector to make the project, assign tasks, and set dependencies. Ten minutes later, I have a fully fleshed-out project. Could I have created it manually quickly or used built-in Asana tools? Yes, but the learning and efficiency gained by taking the extra step and time were well worth it.
We are testing keeping school district contact lists up to date using AI agents managed by Google Agentspace. We don’t have plans to put it into production, but we are researching what it would look like and how it might formulate our AI strategy in the coming year. There are a hundred ways to do this, and we are duplicating work (We have our team update our contacts manually every fall). However, having an AB comparison of how well the agents do, using Agentspace as the orchestrator, and having Google partner to support the work means we will build knowledge and capacity for what is next.
Links
Writing competition https://mark—lawrence.blogspot.com/2025/08/so-is-ai-writing-any-good-part-2.html
Before the Commit https://podcasts.apple.com/bh/podcast/before-the-commit/id1825463283
Dwarkesh Podcast and Blog https://www.dwarkesh.com/
Zak El Fassi https://zakelfassi.com/
Talk and Comment https://talkandcomment.com/
Intelligent Machines https://www.youtube.com/user/ThisWeekInG and https://podcasts.apple.com/us/podcast/intelligent-machines-audio/id326120877
Edtech Insiders https://edtechinsiders.substack.com/
Claude Code https://www.anthropic.com/claude-code
Google Agentspace https://cloud.google.com/products/agentspace
Asana MCP https://developers.asana.com/docs/using-asanas-mcp-server
