What are some effective ways to use AI for personalized student feedback?

Imagine using a tool like Squirrel AI that looks at your quiz answers and tells you exactly where you're doing well and where you might need more practice. For instance, if you're awesome at solving math problems but find word problems tricky, the tool could suggest extra exercises to help you improve. It's like having a study buddy who knows your strengths and helps you with the tough parts. Additionally, there's Knewton, a tool that adapts to your learning style. If you prefer learning through interactive activities rather than reading, Knewton can adjust the content to match your preferences. It's like having a personalized learning journey that suits you perfectly.

Interactive Dialogue and Coaching can be a helpful method: 1. Virtual Tutors: AI-powered chatbots can simulate conversations with students, answering questions, providing explanations, and offering encouragement. 2. Personalized Study Guides: AI can generate customized study guides based on individual knowledge gaps and learning preferences. 3. Progress Tracking and Motivation: AI can track student progress and offer personalized feedback and encouragement to keep them motivated and engaged.

Personally, I suck at feedback! I suck at receiving it, and I suck at giving it. It’s not easy to know what to say, how to say, and when it’s appropriate. No, the timing of it is less likely to benefit from AI. But what to say and how to say, it can definitely be approved by a more empathetic and understanding response. A simple example of this might be to prompt, the GPT with something like “act as an empathetic counselor…“ Then ask it to give a response based on a certain action or assessment that you have deemed needs a response. For most of us, whatever the AI generates is likely to be far superior to what we would think of on the spot. It can analyze thousands of responses, synthesize the best parts, and there you go!

I love to leave myself quick notes about students in WhatsApp using speech-to-text. I'll then copy/paste the notes, copy/paste results or data about the student/ and then add a prompt to GPT like “using as many of my words as possible, give thoughtful feedback from the point of view of a warm/empathetic counselor. Focus on the positives as well as providing encouraging next steps. Give this feedback in 3 paragraphs of no more than 5 sentences.“ I then ask for 2 or 3 different outputs. I use AI to do 80% of the grunt work, personalizing it at the end by combining the ideas from the different outputs and ensuring that the final feedback sounds AND feels, like me!

A great way to improve feedback with the use of AI is to change the tone in which you deliver feedback. I find this really useful for when I know how particular students might react to suggestions of improvement. Sometimes, I find it difficult to provide feedback in a way that is different to the way I would want to receive feedback, so AI tools can really help to give some insight into how rewording can make a huge impact despite getting across the same point.

At times, as educators, we tend to forget one of the rubric or parameter which can be compensated by AI. Generative tech like BARD and GPT ensures the feedback is precise and all rubrics are covered.

AI-powered tools can analyze student work and responses to provide personalized and specific feedback aligned to their needs and abilities. This allows teachers to augment their own feedback, reaching more students in less time. AI feedback helps identify knowledge gaps, misconceptions, optimal next steps, and resources tailored to the individual. It enables students to receive high-quality, targeted, and timely feedback at scale while respecting teacher bandwidth limitations. Implemented ethically, AI feedback facilitates self-directed learning, increases student engagement via agency and ownership, and supports inclusion by meeting each student needs.

AI driven feedback based on the parameters of a rubric can take the feelings and emotions out of grading. It makes it much more objective.

AI can be of great use when it comes to analyse the data of one’s performance. AI can generate graphical representations of one’s scores secured over time and can give you insights about one’s performance in much more communicable way.

In my experience one of the most useful applications of artificial intelligence has been the use of essay grading tools. With the option of determining and customising one’s own rubric, the AI is capable of objectively marking and grading entire papers in a matter of seconds. It should then be obvious to any teacher, lecturer or educational instructor why this is such a great convenience! Besides, the massive time benefit to teachers notwithstanding, it reduces the response time to students exponentially.

Difficulty with any GPT is to know how to prompt it properly. “Prompt engineering“ is becoming a vital skill set in today’s society, much like typing was in past decades. Explore advice from others online about proper prompts, use custom-made GPT’s, and keep a running notepad with prompts that you’ve had success with. This will greatly increase your chances of having future successes! Also, make sure not to put student personal details within the chat as this could be an ethical privacy violation. we don’t know where much of this information goes, but it’s safe to say anything used within the prompt might be visible to others now, or in the future. Happy prompting!

AI has been a very useful tool for report writing in an international setting the ability to communicate with parents in their own language through translation tools has made their transition so much easier.

Now AI may not be able to comprehend everything that we provide and hence the output might be something unwanted and unexpected. At times we might feel that a human would have done a faster job as compared to calibration that might be required by AI.

Despite my outspoken endorsement of AI in education, it’s worth acknowledging that it has its drawbacks. The AI interpretation of certain rubrics or approach towards various topics or concepts are always subject to the AI training model, presenting a host of unprecedented issues. For instance, just the other day I had Google Bard set a personalised test for one of my students, which it did flawlessly. The problem, however, arose when I asked it to provide me with an answer sheet, to which in typical AI fashion, it refused to give me several times, citing ethical concerns. These flaws in the system, however minor, are irritating, and in time will need to be fixed before we can seriously consider implementing AI into mainstream education.

All of these are excellent ways to use AI in the language learning classroom. In most situations, there is one language expert (the teacher) in each classroom of 20-40 students (on a good day). Having students submit writing samples for feedback from AI (with a warning that it may not be perfect) or to have them chat about a particular topic with AI would radically increase the percentage of active student participants.

AI feedback is important for some basic tasks and it will improve your productivity and performance if used in an appropriate manner. No plagiarism is acceptable but if used for learning purposes then its beneficial.

It's important to comprehend that AI is data-driven and “artificial” intelligence, not “real” intelligence, so it should be utilized in sense of data-based analytics. My suggestion is to observe feedback in a hybrid mode, using both artificial intelligence and human intelligence. AI should not be the sole decision maker, it is important to involve teachers and students in the lifecycle too. Taking the pros and cons of technology into account, we can make it effective while ignoring the negative aspects (cons). It's important to embrace change in our fast-paced world and stay ahead of it so let’s be on top of AI acceptance.

Encourage Critical Thinking: Instead of solely relying on AI feedback, teachers should encourage students to critically evaluate the suggestions provided and engage in a deeper understanding of the subject matter. Promote a Growth Mindset: Emphasize the idea that feedback, whether from AI or teachers, is an opportunity for improvement. Encourage a growth mindset to foster resilience and perseverance. Train Students on AI Use: Educate students on how to interpret and utilize AI feedback effectively. This includes understanding its purpose, limitations, and integrating it into their learning process.

AI should never solely replace you as the core feedback provider. I always write my feedback first, unifluenced by AI, before I seek feedback on my feedback. This is a process I would do working as a tutor in Explore Learning. I would write some feedback for student's work and ask my managers for feedback on my feedback creating a loop. However, when managers and human resources are not available, AI can be a great alternative. I find the best results are gained when doing it in this manner as oppose to using AI first for a feedback draft and then you looking over it and tweaking. There is less human element in the latter.

I would say that as a general code of conduct, informing students that you’re going to be making use of AI or any technological tools is a given. While most of us are thrilled to adapt to these ever-changing innovations, not everybody shares that level of optimism surrounding technology and the direction in which it’s heading. Looming concerns over privacy and data protection may invite reservation in some students over whether or not they would like their information processed by and shared and with companies such as ChatGPT, famously a subsidiary of Microsoft. These matters should therefore be dealt with diligently and sensitively.

I was contemplating whether or not to go down this rabbit hole, but as of writing this I’m at liberty to say that Google Bard by far outpaces ChatGPT in almost every aspect regarding productivity, and its application as a teaching tool is no exception. While ChatGPT is much better at dealing with follow-up questions and translating lengthy texts, Bard designs lesson plans, lays out grammar comprehensibly, creates practical and useful examples, sets quizzes and tests in any given format and delivers personalised content far more efficiently and effectively than ChatGPT.

There are no free and paid for customized GPT‘s through third-party software available. Some customized for research, some for writing MCQs for educator assessments, and a host of other purposes. Plus, you can make your own custom GPT within ChatGPT. Now, the latter is more difficult, and takes a lot of tweaking to get the correct answers that you’re looking for. So utilizing one’s created by Programmers and tested for validity might be a good way to go. Just consider that these can break at any time, go through unexpected updates, and we never know for sure how privacy settings work within them. Using them is not forboden, but should be done with caution.