AI Update for K-16 Administrators: More People Need to Step-Up and Take the AI Bull By the Horns
AI capabilities are way beyond what most schools are aware of, and they will transform education and society over the next few years.
TLDR
*Small changes: Apple Vision Pro; Brilliant glasses with embedded bots; Gemini
*Bigger changes are near: Agents that can work autonomously by giving themselves their own tasks, and other agents coming to Microsoft
*Larger changes on the horizon: Robotics and hyper-intelligence
*Misunderstandings schools have: ChatGPT and generative AI are primarily what matters
*Administrators keeping up with the big picture: A book and a report to read
*Taking the bull by the horns: Administrators who work to prepare students for the world that actually exists
*Products useful to administrators
*Student workforce and college preparation
Today’s update is focused on what I think is important for school administrators to understand about AI. I also think it’s useful for everyone to understand, but this post excludes content that I think is less important for them.
Rate of Change
Micro. Last week saw the release of the Apple Vision Pro (AVP) and the announcement of the Brilliant glasses (think of something similar to the AVP but not as sophisticated or expensive, but can do the basics).
Students and faculty who wear these devices, and other similar ones (Meta glasses) in school can access bots and basically make any cell phone bans irrelevant. These are cell phones on steroids that immerse individuals in augmented reality environments behind their prescription lenses without anyone knowing they are engaged with a bot.
Google also released its Gemini Ultra model ($20/month with two free months). It also retired Bard and replaced it with some enhanced features, though not everything that is included in the premium subscription. Some unique features include the ability to adopt different personas for role-playing and exploring writing styles, integration with Google research resources, up-to-date internet access, data upload, and support for specific writing styles.
Some people think all of the above is a really big deal, and they certainly will impact education in terms of how students do school work and an ability to teach them in immersive environments, but they aren’t anything that is unexpected and they are in line with what is expected.
A bit bigger: Agents. We are very close to meaningful AI agents. Basically, agents can be given a task and complete a multi-step process. For example, this one can arrange your Valentine’s Day date. It does everything on the right automatically.
Andrew Ng says these agents can currently work for approximately 30 minutes autonomously, but he expects big improvements this year. I think it's the next big thing we'll see in AI. Incremental improvements in ChatGPT4 vs Gemini Ultra are interesting but relatively trivial developments IMHO.
New advances in CoPilot are coming with two features:
(a) a button that will light up and suggest actions for you to tell it to do (create a PowerPoint);
(b) When you hover over it, it will have a list of the suggested actions for you to click on.
This is basically an agent that isn’t yet empowered to take the multiple steps autonomously (you’d first instruct it to write your report and then when given the option in the next step to have it create the PowerPoint, but you can do all that at once (or you will be able to soon): Generate a research paper with an accompanying PowerPoint.
China’s new AI baby is an Agent that can act on its own by independently giving itself new tasks.
So, like a human teacher, a teaching Tong Tong could realize a human student is struggling and automatically adjust its approach to teaching that student.
Long-term but not far way: Moving toward embodied human-level hyper-intelligence. Current “frontier” AI models have trained on the entirety of the publicly available internet (around 10 trillion words). They are now being fined-tuned on specific, largely text, data, but they still have largely been trained on words.
Now they are starting to be trained on videos, which contain way more data than all the words they have been trained on. As they develop agent abilities. That text and video data, which is now “multimodal” because they are being trained on it at the same time will create a massive amount of additional data that leaves existing sets and training “in the dust.” As machines develop into robots, they will learn to stand up, the same way a human does — from all the text, video, and human interaction they have.
These machines will then share this data with each other and work in a collaborative manner, basically being, at a minimum, the combination of all human knowledge and intelligence, but since they can see cross-human and cross-knowledge, as well as process more data than a human ever could, they will likely be smarter than us (they already hare in many ways).
All the children and young adults in our schools will live in this world and we need to prepare students for a different world than the one they live in.
Misunderstandings Schools Have
It’s just about ChatGPT. There are literally hundreds of thousands of AI models operating at this point, with some small enough to fit on phones and inside eyeglasses.
It’s only about generative AI. There are so many seminars being offered to schools about generative AI and generative AI is important, but there are many more models, many more AI systems, and there are permutations of models and systems. Generating text and images is a tiny subset of what AI can do, especially when put into applications.
Schools need to work on strategies to teach students to live and work in a world where their computers are more intelligent than them (have more knowledge, can learn faster, can apply their knowledge faster, and can do a majority of jobs (or at least job tasks) that humans can, and can decide to take actions on their own.
What to Read to “Keep Up”
Educational administrators should not worry about every AI development, but should, instead focus on the big picture, as those big picture changes will change the entire world and the educational system.
AI and related technologies (robotics, synthetic biology, and brain-computer interfaces) will continue to impact society and the entire educational system over the next 10 years. This impact on the system will be greater than anything that has happened over the last 100 years, including COVID-19, as COVID-19 eventually ended and the disruptive force of these technologies will only continue to develop.
To understand these massive changes, I suggest reading these two works.
Mustafa Suleyman’s The Coming Wave. This book explains all the changes and the context for the changes. It is forward-thinking. Some considered it “hype” when released, but the time for both agents and robotics is far faster than even he anticipated.
Our report. Our continually updated report covers all of the major developments in AI, what they mean for education, and how schools can begin to try to adapt.
If you read the book and the report, you’ll be able to start making informed decisions about how to lead educational adaptations in this world.
HOPEFULLY SOMEONE WILL TAKE THE BULL BY THE HORNS
AI is the bull in the China Shop, redefining the world and the educational system. Students writing a paper with AI is barely a poke in the educational world relative to what is starting to happen (active AI teachers and tutors; AI assessment; AI glasses; immersive learning environments; young students able to start their own business with AI tools; AIs replacing and changing jobs; deep voice and video fakes; intelligence leveling; individualized instruction; interactive and highly intelligent computers; computers that can act autonomously; and more).
No one likely has any experience with this. We’ve never had anything like this in a school before, no matter how much we want to associate it with “edtech,” and there isn’t much in learning theory that assumes people will spend their entire lives working with highly intelligent machines in a shared intelligence framework (in fact, we’ve seen educators who are opposed to AI saying students must demonstrate “their own” knowledge).
Who might have experience with this? Schools that assign each student a tutor who has PhDs in all the content areas the child is learning, an advanced degree in instructional design, and a lot of experience, including working with the child.
This is a whole new world, and it will be essential for educational leaders to step up and guide the educational system forward if it is going to survive. Taking three years to implement important and already agreed-upon changes will leave the system in a state of irrelevancy.
What is going to be required is something like the Committee of Ten, led by Charles W. Eliot of Harvard University and William Harris, U.S. Commissioner of Education, who reset public education in the 20th century to support the move from the agrarian era to the industrial era. The difference between this and the agrarian and industrial eras is small compared to the difference between the first industrial era and Industrialization 5.0 where humans share intelligence, and perhaps even body parts, with machines that are smarter than themselves.
Products useful to administrators
Here I will highlight some apps that are useful to administrators or the administration of the school or district.
Jellypod: This tool allows you to convert your emails and newsletters into a personal podcast. This is a great way to keep your community informed as they go about their daily routines.
HeyGen. Here, you can convert any podcast or web video to many languages using your own voice and expression. Get your announcements out to individuals in your district who speak all languages.
Guide. Download the Guide extension on Chrome. It's an AI-powered tool that records you doing a process and automatically turns it into a video with instructions.
Click on the Guidde extension icon in the top right of your browser and do the task that you want to train your employees to learn.
Click on the extension icon again when you're done.
Go to Guidde’s website and find a recording of the process, along with captions and instructions that are automatically added to the video.
Make any adjustments that you want to the captions and instructions.
Share the recording with your team or customers so they can learn how to do it without you having to explain it over and over again.
Power Point. Take anything you’ve written and have Microsoft Office convert it to a PowerPoint.
GPTs. Build simple GPTs to automate your tasks.
Prepare for Enhancing Learning: Increasingly Interactive Educational Experiences
The realm of education is also witnessing a transformation thanks to deepfake technology. It enhances online learning, making it more engaging and interactive, especially crucial in an era where digital learning is becoming increasingly prevalent. In Shanghai, Associate Professor Jiang Fei utilized an anime version of himself to teach during the pandemic lockdown, noting a substantial increase in student engagement.
As Jiang Fei pointed out, "The enthusiasm of the students in class and the improvement of the quality of homework have made obvious progress." This example highlights how deepfakes can revitalize the educational process, making learning more appealing and effective for students.
Prepare your students for college and the workforce where AI is being used.
College Prep
Recently, a lot was made about a professor at Oxford who required his students to use ChatGPT4 in their essays. While this drew some attention since it’s from Oxford, many professors in the US have started encouraging students to use AI as part of their writing. I certainly don’t mean to suggest that all professors are behind this, and it is obviously controversial, but it is becoming more and more part of college academics, including writing. Preparing students for college means preparing them for this expectation.
The World of Work
Great summary article of the challenges and opportunities Stefan
I am just your average doctor turned science teacher, now coaching top students in Science Olympiad national tournaments. I started reading this and it reminded me of recent "discussions" I have had recently with people telling me I "don't understand" AI, or I probably "disliked computer when they came out", or AI is "no different than a calculator", or I am just a "luddite". I am fairly well read. I got through physics and organic chemistry yet about halfway through the article and I had to stop as it was giving me a headache. If AI is this complicated, how are kids (who are already using to cheat on essays, homework, college applications and pornographic photoshopping) ever learn to differentiate between "good AI" and "bad AI" when I can't grasp the difference? It's me. right?