What is AI feedback?
At Atomi, we use Artificial Intelligence (AI) to provide personalised feedback on our short answer questions. This feedback is based on your answer's accuracy, relevance and depth. The feedback is structured in two sections:
What areas you did well in
And, what areas you can improve on.
This structure is designed to make sure you’re across your strengths and weaknesses in the content, showing you how to move forward in your learning.
What is AI marking?
In addition to AI feedback, our AI will mark or grade some short-answer questions.
When you submit a response, our AI will grade it as either Great, Okay, or Needs work. You’ll only see this grade if the AI is confident: so if there’s any doubt on our end, you’ll be asked to self-mark your response as usual.
How do the marks and feedback work?
Our AI system uses existing data to determine how well you’ve answered a question. The data used includes:
Previous question responses
The question information (like the prompt, the sample answer and the marking guidelines)
And, the instructions from our AI specialists who help direct the AI.
This means that your marks and feedback aren’t just a guess. Instead, the model uses those answers, instructions and pieces of information to make an informed judgment.
When you submit a response, the system uses natural language processing (NLP) to understand and evaluate the text based on these factors. “Natural language processing” means the system is based on text and language, so it:
Interprets all the text-based instructions and data we give it
Measures your response against this data
And, delivers feedback back to you.
How accurate are the marks and feedback?
Atomi’s AI is trained, tested and evaluated based on thousands of Atomi questions and actual student responses. We maintain tight supervision by our in-house machine learning experts, academic content creators and high school teachers. This process enables Atomi’s AI to reliably deliver feedback that accelerates learning and meets users' expectations.
Plus, our AI marking function only activates when the model is confident in its accuracy. This means if there’s any doubt, it will default back to self-marking, allowing you to choose how well you did based on the feedback and your knowledge.
But, all AI systems have their limits—at the end of the day, AI is a tool, not a replacement for educators. We acknowledge that our model might have limitations, including:
The potential for misunderstanding highly complex or creative student answers.
Limited context awareness beyond the immediate question or lesson.
The potential that models may provide biased, inaccurate, out-of-date, or otherwise contested information or solutions.
What if I disagree with the marks or feedback?
Every time you’re given marks or feedback from our AI, you’ll have the chance to provide feedback. This could be in one of two forms:
In questions where you receive feedback and no grade, you can give a “thumbs up” or a “thumbs down” to let us know how helpful the feedback was.
In questions where you receive feedback and a grade, you can “Let us know if we got it wrong”. This will open a form where you can write feedback to be reviewed by our team and maybe your teacher.
If you want to provide a more detailed review on how the AI works with your learning, you can contact our support team at support@getatomi.com.
Need more information on security? Check out our page on AI marking and feedback: technical and privacy support.