This Post Was Also Written By AI

Implications & Opportunities for Teaching & Learning

In the two months since my last post on this topic, the AI-powered world has gone WILD. Large Language Model (LLM’s) like GPT3 are freely available and easily accessible, and example outputs from OpenAI’s ChatGPT are all over Twitter. AI Art is exploding, HuggingFace developer community is sharing new models and ideas almost daily, and the ethical and privacy questions are coming.

Linked Posts:

• This Post Was Written By AI
• Thread of experiments with AI (scroll backwards)
• (If You) USEME-AI model for adapting to AI in schools

One of the key concerns educators might have is “how will AI affect student work?”, so let’s find out…

Quick Examples in GPT3

Here is an example page of prompts and output types I’ve made over the last couple of days.

It uses Craft note-taking app‘s GPT3-powered Assistant feature. Each output took under 30seconds, and you can see it type as it generates, with some pretty impressive results, including:

• Compare and contrast
• Suggest and explain
• Synthesising ideas
• Step-by-step explanation
• Essay prompt, rubric and exemplar creation
• Multiple-choice quiz creation and answers
• Creative Writing (sonnet, script, short story)
• Generating potential definitions
• Summarizing academic text
• Recipes and unit conversion

It can’t yet handle mathematical notation or diagrams (though does give working LaTex), and is not successful with including linked, current and reliable references. Yet. However, we can’t get access to OpenAI in China (even with a VPN and virtual phone number), and some of the outputs coming from ChatGPT are much more impressive, including coding and reasoning (to an extent).

If you don’t want to make a Craft account and can’t access ChatGPT, you can get similar results online for free with serviceai.tech.

So what can we do about or with it in schools?

This is going to be a conundrum for academic integrity in schools, for sure. Currently undetectable by plagiarism checkers, some of the greatest risks in integrity might be in assignments with an ‘invisible middle’ – where the assignment is set, there is a gap where the teacher is not seeing the work being done, and then it is handed in. GPT3 can easily handle content in the middle-high school range and can adjust for voice, and formatting, with the right prompts.

Thinking through what this might mean for learning & assessment design, here are a few of my suggestions:

• Leaning-in to using AI as a research-starter tool, teaching students to evaluate and explain the outputs created. 
• Model ethical use of information, images and AI.
• Host socratic seminars or discussions where students explain their work and are trained to ask and respond to meaningful questions. 
• Create authentic assignments from which personal connections must be made, and which require references and citations of quality resources. 
• Creating assignments that bring current news (before the model’s training data) as case studies to be applied to the concepts being studied. 
• Using pen-and-paper assessment for knowledge. 
• Using AI prompts to generate interesting question and task options for students to take action on in ‘real time’

Then I thought, “what would the Assistant suggest we do?”, so I asked and this is what it came up with:

Prompt: Suggest and explain ten things schools can do to adapt to a world where AI is used in assignments and teachers might not know if the work is authentically the students’ own. 

1. Establish clear rules for students about the use of AI and explicit consequences for breaching these rules.
2. Train students in how to use AI correctly and for what purpose.
3. Use AI to supplement and not replace teacher-student interactions.
4. Create higher-order tasks that require critical thinking and creativity, which AI is not yet capable of.
5. Encourage peer-critique and review of assignments so that students can help hold each other accountable.
6. Provide detailed instructions and rubrics for assignments so that the work is meaningful and original.
7. Facilitate creative projects such as art, music, and robotics, where AI is unlikely to be used.
8. Encourage students to produce multimedia content and stories (visual, audio, or animations) as a form of assessment.
9. Use AI tools to help automate the marking process, thereby freeing up teacher time to provide more personalized feedback and support.
10. Work with parents to ensure that students understand the importance of using AI responsibly and ethically.

It’s becoming clear that the successful knowledge workers of the future will be “AI augmented” and those who do not know how to make good use of these tools could be left behind.

So, yeah, an AI could have written this blog post too.

AI2041 is coming, perhaps sooner than we think.

Shared on LinkedIN here, if you want to continue the conversation.

When I asked the GPT3 Assistant how schools might adapt to the AI world, this is what it came up with: pic.twitter.com/jZhC4DcSvG

— Stephen 🌏 Taylor (@sjtylr) December 8, 2022

New Post: (If You) USEME-AI (Draft1.0) for adapting to #AI use in schools.
💻 Post: https://t.co/ZedVvhfcJd

Could Creating a Culture of Thinking & Purposeful Learning mitigate implications & empower opportunities for meaningful applications & action? #EdTech #EdChat pic.twitter.com/PxcDJxA2Ke

— Stephen 🌏 Taylor (@sjtylr) December 10, 2022