21st Century Teaching: using AI in the classroom

New technologies are often met with high amounts of skepticism and rapid integration: smart phones were seen as a fad, but quickly became the dominant form of communication, social media was seen as a young person’s medium but is now one of the largest sectors of the tech industry. AI is no different. Many people were hesitant to accept this new technology, sparking fears from movies like 2001: A Space Odyssey and Wall-E where AI systems turned evil. However, in just two years since the launch of Chat GPT, new AI tools seem to spring up everyday. For teachers and students, this explosion has presented new challenges and opportunities. One challenge that is perhaps most paramount is bias.

Addressing bias in AI

There are many ways for bias to creep into the large language models that many AI tools are built off of. Below I address the different instances of bias and how we can make sure to avoid them in the classroom:

1. Bias present in the data sets of large language models
   • AI chat bots are built using data collected from a variety of sources including human and societal sources. There are engrained biases in our society: gender, racial, and ethnic biases will be present in the data that is fed to these models. An example includes programs that are trained on criminal records would reflect the disproportionate number of arrests that people of color experience versus white people, and reproduce that result in its output.
   • Edutopia recommends one way to counteract this bias is to move beyond some of the bigger LLMs like Chat GPT and use models that have been trained on more diverse data sets such as Latimore.ai. This blog post dives into more ideas such as teacher training programs addressing data analysis in their programs and being proactive and comprehensive in addressing bias before presenting to vulnerable student populations
   • This ties in with the TCE threshold concept teaching is not neutral/teaching is political addressing bias is not a new role for the classroom teacher. Textbooks, standardized test, and other educational materials have been rife with bias for years. Especially in the context of science, black and brown voices have been marginalized and silenced in materials presented by many curriculum produces. Addressing bias in AI should be treated just the same, it should be seen as equally important as the content being taught.
2. Treating AI as an authority
   • It is one thing to acknowledge that AI has bias engrained in its program, but it is another to blindly trust AI. Teaching students how to critique primary and secondary sources, or a news article about a scientific concept are important skills taught in many lessons throughout the school year. A essential prat of that lesson is just because somebody presents themselves as an authority does not mean that they are the final word on the matter. The same critique should be done on AI, just because the chatbot says its true does not mean that it is.
   • This aligns with the TCE threshold both teachers and students have empowerment/agency. One way for students and teachers to express their agency is to not only take back some of their time by using AI to reduce their work load, but it is also an expression of agency to critique the content that they are presented with! We must take ownership of our learning and teaching by deciding when and how we use AI in our classrooms, it is our decision as teachers when we want AI assistance, and it is our role to help our students understand where the dangers of using AI will be present.

Magic School AI

This week I used Magic School AI’s writing feed back tool with my students. Their task was to write a letter to the Superintendent of a Park about whether Kudzu (an invasive species found throughout the southeast of America) should still be classified as non-native and therefore eradicated. As a science teacher I have not always been the best at providing feedback on my students writing, but it is a building wide expectation that we provide opportunities for our students to showcase their writing across all subject.

I was very impressed by the ability to set up a “classroom” within Magic School for my students to create an account. This made it possible for me to see everything my students did while they were using the AI tool. I posted a screen shot of the writing input and feedback one of my students got on their writing.

I was very happy with the outcome of using the tool in my Biology class. It saved me a lot of time and provided immediate feedback to my students which is an important part of their growth. However, I was still able to provide more specific feedback on my own that went beyond what the AI tool could provide. One thing that I would improve on next time is providing a rubric for my students so that they could include it in the writing feedback tool so that they could get more specific feedback on the parts that they were being graded on. Overall, I think that more tools like this should be the focus of using AI in the classroom.

Pros and Cons of AI in the Classroom

There are lots of ways to think about AI in educational spaces but it is important to address both the benefits and draw backs to using AI.

Pros:

• Individualized Instruction
  • Like I did this week with my students, using AI can provide individualized feedback and instruction for students in a way that has never been easier. With automated tools like those from Magic School, teachers can individualize the assignments for different reading levels or abilities, without doing much more than pressing a few buttons
• Reclaiming Time
  • I see one of the biggest impacts of AI is the time it can save for teachers. Lesson planning, grading, parent communications and more can all be streamlined or eliminated almost entirely through the use of AI. I know that me and other teachers I talk to often complain about the mundane tasks that take away from our time doing the thing we love: teaching! AI tools can give us some of that time back so that we can recharge and be ready to serve our students.

Cons:

• Bias in the model
  • I’ve already spoke extensively about bias in large language models, however, it can’t be overstated the importance of addressing bias. Our students deserve to be protected from harmful bias that they could be exposed to by using AI. Bias from AI could also lead to reinforcing harmful stereotypes that some students already believe and are hearing at home. Addressing bias has to be at the forefront of using AI.
• Error/Incorrect information
  • We are still early in the in the roll out of many of the most popular tools, almost all of which are less than 5 years old! They are still getting things wrong, they are not perfect, and we must keep that in mind as our students use these tools. It is important to show them how to determine if the information they are receiving.

For more one the pros and cons of using AI in educational spaces check out this blog! As well as my latest tweet!

Being a teacher in the 21st century requires us to meet new technologies head on! Check out my new blog post on AI where I address bias and the pros and cons. I even used AI in my class this week! @NSTA @edutopia https://t.co/Aq7qyUMzhF

— Mr. Turner (@MrTurne82324326) October 25, 2024