SAIL: It's a Leadership Thing, All the Announcements, Environmental impact
May 22, 2025
Welcome to Sensemaking, AI, and Learning (SAIL). I focus on AI’s impact on higher education.
AI and Education:
What should Boards, Presidents, and Provosts know? Since the late 1990’s (when I first encountered online learning - at that point, mainly html pages on college sites), I’ve been focused on how technology impacts systems of learning. At each stage - open education, web 2.0, learning analytics, MOOCs, AI & education - the feeling has been one of the system must change. This a recording of a recent talk on exactly that topic, but directed to senior university leaders. This is a useful framing was well - sandboxing rather than trying to only build within (where existing culture can starve innovations or prematurely force legacy practices to shape emerging capabilities)
The heavy agentic focus of AI currently is producing frameworks and concept papers. Galileo offers their own “AI agents introduce unique evaluation and testing challenges that few engineering teams are prepared to address. These systems operate on non-deterministic paths, capable of solving the same problem in multiple ways.” This matters in higher education because the education process isn’t singular (like an agent that checks inventory for a previous purchase) - it’s complex and multifaceted. We can expect agents that teach, assess, create content, and direct students to resources. That will require a multi-agent approach and orchestration will be key. That’s why Google, Meta, OpenAI, and others are creating agent frameworks.
Google udpdated LearnLM. (see more about Google announcements this week below): “Learning science also shows that providing information in multiple modalities can help enhance learning, and with Gemini’s multimodality, there are more ways than ever to remix info”
AI in General:
It was a big week for big announcements in AI.
Here’s what Google dropped. Have a look at the first image to see the pace of development over the last year. Project Astra, announced late 2024, looks particularly promising for multimodal interaction with the world. Veo 3 is impressive - learning designers and creatives can create quality video outputs for a fraction of the traditional development costs. Oh, and Android XR so AR/glasses move toward consumer accessible domain.
Microsoft wants to own agents. Their developer (well, Build) conference this week planted that flag: “We envision a world in which agents operate across individual, organizational, team and end-to-end business contexts. This emerging vision of the internet is an open agentic web, where AI agents make decisions and perform tasks on behalf of users or organizations.”
OpenAI is solidifying its role as a consumer technology. Jony Ive, of “iPhone design” fame, has joined to lead new device development. A self-indulgent and AI-feeling video was dropped as part of the announcement.
Anthropic released Claude 4 today. And then dropped this somewhat ominous post. Basically, this model might be getting more dangerous. But they’re not sure. But for now, a headline will do.
It seems so long ago, but seven days ago, OpenAI dropped their software engineering agent, Codex.
Mistral (the one odd non-USA or China AI company) dropped its own agentic SWE, Devstral.
MIT Tech Review posted a few articles on the energy and environmental impact of AI (and held a webinar). Video is the killer in terms of energy use. Two articles to note: