EnlightenAI: Your Virtual TA

In today’s episode, we'll explore how EnlightenAI can assist in evaluating student writing and providing valuable formative feedback. Visit AVID Open Access to learn more.

Transcript

Paul Beckermann 0:00 Welcome to Tech Talk for Teachers. I'm your host, Paul Beckermann.

Transition Music with Rena's Children 0:05 Check it out. Check it out. Check it out. What's in the toolkit? Check it out.

Paul Beckermann 0:16 The topic of today's episode is Enlighten AI: your virtual TA. If there's one thing that teachers never have enough of, it's time. There are only so many minutes in the day, and it's barely enough to get the basic needs met, design a lesson for the next day, grade student work, attend meetings, and answer student questions between classes.

Wouldn't it be great if every teacher had a personal TA or teaching assistant to help with some of the more mundane and time-consuming tasks? Well, in today's episode, I'm going to introduce you to a new AI tool that just might serve that purpose. It's definitely not going to replace you, but it can lighten the load and help give you faster, more timely writing feedback to students. It's a program called Enlighten AI. At its core, Enlighten AI helps grade student writing and provides both students and teachers with timely, targeted feedback. Let me walk you through a summary of how it works and some of the key highlights.

The first thing that you would do is design the writing assignment and grading rubric, just as you would do in an offline task. These details would then be uploaded or entered into Enlighten AI. From there, the assignment is assigned to students in the classes that you've created. The students then complete and submit that work.

Since many teachers are already using a learning management system to help facilitate this give-and-take process, Enlighten AI has integrated some of the most popular tools into its system. For instance, it offers direct compatibility with tools such as Google Classroom, Google Docs, Google Sheets, PDFs, Canvas, and Illuminate. Once you have the student submissions back in your hands, now you train the AI on how you want them to be graded. This is the part that's really unique and impactful. You, the human in the equation, stay in the loop, providing key training and input to help the AI do a better job grading.

It's sort of what you would do with a human TA; you would train that TA how to grade the assignments that you've given so that the grading is consistent with what you would do yourself. To do this in Enlighten AI, you grade five samples of student writing. When you open a student's scoring screen, the AI will automatically attempt to score it based on what it already thinks is the right answer, and it will add comments based on the grading option that you've chosen.

Then you assess the student work and change any of the AI grading outputs in a way that you think better matches how that student performed. This includes adjusting the points given on the rubric scale or changing the personalized comments that are included with the rubric. If you agree with the AI or like the comments, you can keep them, but you also have the ability to override those. As you adjust and accept these choices, the AI learns from your preferences.

It takes about five samples for the AI to become really good at assessing the assignment and providing personalized feedback. Now you might be asking, "Can the AI really grade as effectively and accurately as I can?" Well, that's a question I had too. According to a study published in Educational Psychology, the answer is yes, for the most part.

The study is titled, "How does artificial intelligence compare to human feedback? A meta-analysis of performance, feedback perception, and learning disposition." Drawing on 41 studies of 4,813 students, the findings reveal no statistically significant difference in learning performance between students who received AI-generated feedback and those who received human-provided feedback. In other words, in terms of how students perceived that feedback and how it impacted their performance, there was virtually no difference. Now, that doesn't mean that the teacher can be replaced or that there's no value in the teacher.

On the contrary, the study in no way encourages that. It found that the human teacher was still much better at providing empathy to the student and understanding each student's unique context and circumstances. Both of these are invaluable in the classroom. Beyond that, it's still the teacher that is training the AI how to score the work. Again, it's sort of having a TA in the classroom.

In this light, the study recommends a hybrid approach: leverage the efficiencies of AI but also the personal touch of the teacher. So, going back to Enlighten AI: once you have trained the AI, you can now have it grade the remaining papers and provide automated feedback. It will assign scores on the rubric and provide written comments regarding the strengths and weaknesses of the student work. It will base both the comments and the scores on the training you provided.

If you have multiple sections of the same class, you only need to train the AI once, and then you can apply that training to the other sections as well. This is a step that will likely save you hours of work. However, it is important to remember that you are still the ultimate classroom expert, and you can and should go through those results to confirm that you agree, especially if it's a summative assessment. If it's formative, you can possibly let the AI take on a little heavier lift.

In addition to receiving individual feedback, you'll also get a summary for the whole class, letting you know the average scores as well as trends the AI is seeing. For instance, it might report that students are excelling at using textual evidence and examples to support their reasoning, but stylistically they struggle with transitions and coherence. This is valuable data and can help you design the next lesson. And believe it or not, there's a built-in functionality for this as well.

You can use the built-in AI to help you design or brainstorm follow-up lessons based on the assessment scores and comments for your specific classroom. Of course, all of these are editable, and you can iterate the output with a follow-up conversation in the built-in generative AI chatbot. All the way through the process, you are in control.

Now, because the rubric assignments and training can be shared with other teachers and other sections that you're teaching, this opens up another potential use. Enlighten AI can be used as a common assessment tool that pushes out the same assessment from a central location to multiple classrooms. This might be your central office leadership, a PLC chairperson, or somebody in building leadership. By doing it this way, teachers can assign a common writing task to all students, and the work is scored by a common rubric shaped by the same training data. This helps to ensure that all student work is graded similarly and fairly.

These results are then aggregated and able to be used for analysis. The Enlighten AI company has conducted tests to ensure the accuracy of this approach. They've taken human-graded writing and then had the AI grade the same submissions and compared them. The differences were very minimal. In fact, in one demonstration done during a live webinar, the results aligned 100%; the human and the AI scores were the same.

Again, it is important to keep the human in the loop. Based on the research and trials, it appears that Enlighten AI could be a huge time saver, especially in the formative phase of writing. Speaking of formative learning, Gautam Thapar, the founder of Enlighten AI, shared that students can be allowed to revise their writing based on the feedback and assessment they've received. As they change their writing and resubmit it, the scores change. This means that students get real-time updates regarding how their revisions impact their scores.

Gautam shared that this has been immensely motivating for students. As a bonus, there are additional AI tools integrated into the teacher dashboard with options such as a lesson plan builder, text generator, exemplar and non-exemplar creators, worksheet creator, and multiple-choice quiz. You could complete these tasks with your favorite generative AI chatbot as well, but it's nice to have them readily available right in the workspace.

As for pricing, Enlighten AI is currently free to use. If districts want to set up Clever integration, there is a fee for that. Eventually, there will be a paid tier of Enlighten AI, but the founder has pledged that there will always be a robust free plan available to teachers. While I'm sure we'll see more products like this emerge, Enlighten AI appears to be a leader in AI writing assessment at this time.

If you're looking to save time grading, gain quick insights into class writing trends, or develop data-driven teaching materials based on your own classroom performance—or perhaps you're looking for ways to give students more immediate feedback on their writing—then it might be worth your time to explore Enlighten AI and its possibilities. You never know; it could be the tool that gives you back some of that valuable time.

To learn more about today's topic and explore other free resources, visit avidopenaccess.org. Specifically, I encourage you to check out the article collection, "AI in the K-12 Classroom." And, of course, be sure to join Rena Winston and me every Wednesday for our full-length podcast, Unpacking Education, where we're joined by exceptional guests and explore education topics that are important to you. Thanks for listening. Take care, and thanks for all you do. You make a difference.