AI in Education

Update to this Publication

Thanks for checking out our AI Publication. Since publishing in March, announcements and advancements continue to occur at a breakneck speed.

Keep track of AI developments and resources that intersect education on our updates page.

View Updates

Introduction

For the past decade, we’ve been covering advances in artificial intelligence and machine learning, sounding the alarm that it’s not if it’s when and it’s not when… it’s now. Over the last few years, the news cycle appears to be in full agreement with us. This publication highlights trends and developments in artificial intelligence that are shaping teaching and learning. 

We know a few things to be true: 

  • Every day-to-day, rule-based procedure is able to be automated
  • Every sector is growing in complexity 
  • More work is being done by diverse teams on new problems with smart machines
  • Technological innovation is occurring at an exponential rate

Exponential Growth

Tim Urban of Wait But Why explains that “In order to think about the future correctly, you need to imagine things moving at a much faster rate than they’re moving now.”

View Wait But Why

AI is a growing web of related technologies that, given ubiquitous use, broke through to the popular press in 2016. When Google’s DeepMind beat the world champion Go player in March and self-driving cars showed up in Pittsburgh in September, it became obvious that this new cluster of technologies was moving fast and had broad implications. In late 2022 and early 2023, OpenAI started an avalanche of AI announcements with the public beta version of ChatGPT, rapidly followed by GPT-4 and numerous other iterations of Natural Language Processing tools. Everyone was amazed at the believability of these products. The future arrived. 

So what does this mean for us? “Can we adapt fast enough?” That’s the big question. Harvard’s Erik Brynjolfsson thinks that “they already are much smarter than us at many things”.

Homo techne could be translated as “the technological human” or “the human defined by technology.” In using this term, you are arguing that our ability to create and use tools, machines, and other technologies.

Impromptu

What does this have to do with teaching and learning?

We’d be lying if we said anything other than “everything”. At its core, Artificial Intelligence will cause us to question the nature of intelligence which will not only change what we measure but also what we value.

A few of the most notable and urgent impacts are: every field is computational and dynamic and AI competes for attention, thereby requiring renewed values agreements and additional community conversations. 

The education sector must respond by doing some, or all, of the following: 

  • Integrate design thinking & data science into K-12 curriculum.
  • Build, measure and develop human skills: social-emotional, agency, creativity, and entrepreneurial mindset .
  • Create community agreements: a set of agreed upon priorities, values, skills.
  • Teach/practice digital discernment/ethics and renew a commitment to digital literacy for all learners. 

Town Hall: AI in Education

In 2021, we hosted a town hall to highlight a number of the innovations and companies mentioned in this paper. Smaller segments of this video will appear throughout the paper, but you can watch the whole thing here.

My definition of Homo techne assumes that every individual human, sapiens or ante-sapiens, with or without an AI copilot, makes choices. Those choices, in the aggregate, shape what is experienced by each succeeding generation.

Impromptu

What does AI think about AI’s impact on education?

In his recent book Impromptu, Reid Hoffman asked GPT4 how AI could support students around the world in educational pursuits. Here is the exchange: 

Reid: GPT4, can you write the next 300 words from your perspective as a large language model and describe how you could extend the Bridge/NewGlobe tablets into a tool to help tens of millions of students throughout Africa and Asia? 

GPT4: As a large language model, I can help by enhancing the quality, personalization, and scalability of their learning content. Here is how I would do it: 

First, I would use natural language processing and computer vision to analyze the existing lesson scripts, test scores, teacher feedback, and student performance data from Bridge/NewGlobe schools. I would identify the key learning objectives, concepts, skills, and assessments for each grade level and subject, as well as the common misconceptions, gaps, and difficulties that students face. I would also detect the linguistic, cultural, and contextual variations that affect the relevance and appropriateness of the content for different regions and settings. 

Second, I would use natural language generation and computer graphics to create new and improved lesson scripts, activities, exercises, and multimedia resources that are aligned with the learning objectives and adapted to the local needs and preferences of the students and teachers. I would use techniques such as question generation, answer extraction, feedback provision, and adaptive sequencing to make the content more interactive, responsive, and personalized. 

Third, I would use natural language understanding and speech recognition to monitor and support the delivery and use of the content by the teachers and students. I would also use techniques such as speech-to-text, text-to-speech, and machine translation to enable the users to communicate with the content and with each other in their preferred languages and modes.

How AI Works

This is the intro video to a series called AI for Educators from Teach AI. This course is a free, foundational online series for any teacher and educator interested in the groundbreaking world of artificial intelligence (AI) and its transformative potential in education. Partners Code.org, ETS, ISTE and Khan Academy are offering engaging sessions with renowned experts that will demystify AI, explore responsible implementation, address bias, and showcase how AI-powered learning can revolutionize student outcomes.

View Series

Definitions

Adaptive Learning

Changing the presentation of tasks to a learner (e.g. adjusting sequence, difficulty, type of help, etc.) based on the learner’s response to prior tasks.

Algorithm

“In data science, an algorithm is a sequence of statistical processing steps. In machine learning, algorithms are ‘trained’ to find patterns and features in massive amounts of data in order to make decisions and predictions based on new data. The better the algorithm, the more accurate the decisions and predictions will become as it processes more data.” – IBM

Artificial Intelligence

“Technically, AI is considered a branch of computer science and a set of computational technologies. It functions in a manner similar to humans in that it can sense, learn, reason, and then take action; however, AI is not sentient and is simply producing the highest probability answers based on the model data from which it is derived. AI consists of algorithms, rules, and datasets that operate within a specific domain.”

Neural Networks

In AI, these artificial neural networks are software analogues to the neurons and synapses in a brain’s nerve system, receiving, combining and forwarding signals from one node to the next, coordinating a system’s response to inputs to produce an output or decision.

Machine Learning

Machine Learning refers to the process of training AI by developing algorithmic responses based on a set of programmed rules within the context of a dataset. It is the technology that drives tools like adaptive learning platforms and automatic essay graders. 

Deep Learning

A technology that uses inter-connected sets of algorithms similar to a neural network to look for and recognize complex patterns. As a result of Deep Learning, AI can classify images and detect objects, such as with facial recognition. It can also recognize speech patterns and use Natural Language Processing – the technology that drives voice assistants like Siri or Alexa, automatic translation services, and virtual support agents.” – COSN Report

“Deep learning is a subset of machine learning (all deep learning is machine learning, but not all machine learning is deep learning). Deep learning algorithms define an artificial neural network that is designed to learn the way the human brain learns. Deep learning models require large amounts of data that pass through multiple layers of calculations, applying weights and biases in each successive layer to continually adjust and improve the outcomes.

Deep learning models are typically unsupervised or semi-supervised. Reinforcement learning models can also be deep learning models. Certain types of deep learning models—including convolutional neural networks (CNNs) and recurrent neural networks (RNNs)—are driving progress in areas such as computer vision, natural language processing (including speech recognition), and self-driving cars.” IBM

UX/UI

User-interface/user-experience. The acronym refers to the design and description of the way a person interacts with the technology and what that experience is like for the user relative to other approaches.

Next Page

How did we get here?

Artificial Intelligence has been a part of the common lexicon for decades all leading up to this moment.