Pattern Recognition: How can AI support learning and are we even organized for that?

1. Do tons of research 2. Redesign orgs 3. Profit. (IFYKYK)

“We have no idea, now, of who or what the inhabitants of our future might be. In that sense, we have no future. Not in the sense that our grandparents had a future, or thought they did. Fully imagined cultural futures were the luxury of another day, one in which 'now' was of some greater duration. For us, of course, things can change so abruptly, so violently, so profoundly, that futures like our grandparents' have insufficient 'now' to stand on. We have no future because our present is too volatile. ... We have only risk management. The spinning of the given moment's scenarios. Pattern recognition” ― William Gibson, Pattern Recognition

Just doesn’t get better than William Gibson. Probably helps that he describes so well, the primary way that I think. I don’t think that’s unusual - humans are great pattern matchers - but I think I’m hard over on that side. For me, this manifests in my browser tabs. I use two browsers - let’s call this one my research browser.

Lots of pinned tabs and lots of tabs you’ll see in the next Link Roundup but there’s another class of tabs…the ones I keep open because I want to think about them more or because I see a number of them converging on a topic. That’s where we are now.

The first link/story that stuck with me for this issue was Dr Philippa Hardman’s excellent review of a Harvard study on the impact of AI tutoring on rates of learning, its methodology and the potential applicability of the results. This is my fav quote “Perhaps we need to start asking not just how AI can make 1:1 tutoring better and more scalable, but how it might enable, enhance and scale new and alternative approaches to teaching and learning.” Amen.

AI Tutors Double Rates of Learning in Less Learning Time: Link to the pre-publication, not yet peer-reviewed Harvard study - from the abstract: “Here we report a randomized, controlled study measuring college students’ learning and their perceptions when content is presented through an AI-powered tutor compared with an active learning class. The AI tutor was developed with the same pedagogical best practices as the lectures. We find that students learn more than twice as much in less time when using an AI tutor, compared with the active learning class. They also feel more engaged and more motivated. These findings offer empirical evidence for the efficacy of a widely accessible AI-powered pedagogy in significantly enhancing learning outcomes, presenting a compelling case for its broad adoption in learning environments.”

So we are starting to get data, I think importantly from a study of college students, on the potential impact of AI tutors. This has been a dream of this industry, for as long as I can remember. (If you haven’t read Diamond Age, you really should.)

To be clear, there is a lot of research to be done here not just to look at impacts but assuming those land on the positive side (and that’s still a broad assumption at this point), but how to optimize for those impacts - work on UX/UI for example. I also talk a great deal (some would say ad nauseum) about how L&D teams need to think about what activities they can engage in that are higher up the value chain and harder for AI to automate. I think there is a similar challenge here - if you are a teacher, professor, or instructor, and you are joined by a teaching assistant and a research assistant who never tire and never forget anything, what would you have them do? What other things could you do?

There’s another chunk of research to be done too that I think will actually be more difficult to implement. The image below is from a recent paper that shares results from a10-month ethnographic study on how to optimize an org chart with AI in mind. Read the whole study certainly but this point leaps out - “In our study, the algorithms couldn’t fully optimize because the org chart kept decision-making locked in silos. Once those constraints were lifted, AI delivered far better results—spotting trends and opportunities no single team could see on its own.”

That would mean that we can’t get to a place that gives us the chance of optimizing the ROI of the investment in AI, unless we reconfigure our org charts to recognize these new sources. Will this be the new - we need to stop printing out our emails? So now all we have to do is conduct more research on the efficacy of AI in tutoring, do some org design heavy lifting to optimize our internal structures, and so on. So where to start?

I need to add a disclaimer to these next two pieces. When I see implementations of technology that are interesting, I’m typically not looking at the specific company behind the tech or maybe even the specific tech itself but rather at the features introduced and the dynamics those features enable. I also look at these developments in a corporate training light.

In Diamond Age, a young girl from lower social rungs, gets access to an advanced tutor known as the “Primer.” (Little Amazon trivia - the original code name for the Kindle was Fiona - after a character in Diamond Age.)

So now I see the next story and I think back to every new job I’ve ever started and how impactful something like this could be to a new employee or someone in a new role. > >

What Is Google Learn About? How to Use the AI Learning Companion: “So, instead of the standard conversational AI experience with tools like ChatGPT and Gemini, you’ll be able to get a more textbook-style response, complete with images, graphs, interactive lists, and breakout text boxes with labels like “Why it matters” and “Build your vocab.” You can also explore related content and view suggested topics for your given query.” More on Google Learn About - Google’s AI ‘learning companion’ takes chatbot answers a step further: “grounded in educational research and tailored to how people learn.”

Doesn’t that sound like a better experience than hearing “its on the wiki” or “just look it up”? The other piece is that this is accretive. It will get better the longer its implemented and the more information it has access to. We can finally hit that organizational mark of actually being a year smarter at the end of year instead of just a year older. What if all the best practices and lessons learned were at your fingertips? Could you perform better in less time? Now the tough question - what happens to L&D teams if this gets implemented? How will value be generated?

This last one has similar dynamics with new modalities as well. I’ve used Notebook to create an audio version of one of my newsletters. It’s crazy how human it sounds. Take a listen. Now think if you could interrupt it and ask it questions.

Google's NotebookLM Goes Pro: Ken Yeung has been doing some excellent writing on all things AI and he wrote the following about Notebook > > “With NotebookLM Plus, subscribers can now manage up to 500 notebooks and 300 sources per notebook, significantly increasing from the previous limit of 100 notebooks and 50 sources. Additionally, Google has boosted daily usage limits, allowing 500 chat queries and 20 audio generations per day, up from 50 and just three audio generations.” And “To that end, the company is releasing, in beta, a way to interact with Audio Overviews. First teased at this year’s Google I/O, users can interject through an Audio Overview they created, and the hosts will respond accordingly. “It’s like you’re passing a note to them in the middle of the recording of the podcast,” Johnson states.” More on that ability to interrupt the audio overview - Google Now Lets You Talk to NotebookLM's AI-Generated Podcasters: This quote got me “Google says it's like having "a personal tutor" who listens to you and responds directly.”

Crazy right? Org design. AI tutors. Organized institutional knowledge that you can query as easy as asking the person one desk over. This pattern is still forming but some outlines are coming into view. I keep thinking of Amara’s Law.