What is AI?
AI (artificial intelligence) is a broad term for the use of algorithmic technology to perform tasks that would typically require a human’s logic and cognition to complete! Packbacks AI incorporates Machine Learning which is a subset of Artificial Intelligence where the algorithms used are “self improving” by using data, rather than by an engineer updating its code. The great thing about Packback is that our AI uses a combination of many algorithms, some of which are machine learning based (self improving) and some of which are coded based on logical rule sets!
How is AI used on Packback?
In Feedback and Scoring
Our AI system now gives students real-time actionable feedback in four categories and these categories exist to help students improve their scores based out of 100 points. The four categories are: Curiosity, Credibility, Communication and Convention! Our AI looks at a students text in real-time and provides Instant Feedback with a warning to allow them to edit their post before it's ever flagged! Students will be able to apply this feedback from each category to their work in order to create the best possible question or response in the moment. This helps to develop a healthier relationship to feedback that is not punitive towards the student.
For more detailed information on how Packbacks AI is used in instant Feedback please check out our article here: Instant Feedback and Curiosity Scoring
In Featuring
Every week, Packback’s AI will look at recent posts made in a community and identify the highest quality posts for that week, which will then be sent to professors for review. A professor can choose to feature certain posts from the list, or dismiss all the recommendations for the week. If no action is taken, posts are automatically featured and our moderation team will review the posts to ensure they are of high quality, un-featuring those that are not.
What causes a post to be recommended to be featured?
Posts that are recommended to be featured will be among the top quality posts in that community for the week as judged by our AI. In addition, they must not be moderated or have any open flags, be made by a student, and meet certain other quality criteria. Our algorithms get better over time, by learning what kinds of posts are featured and what kinds are not.
Why are posts autofeatured?
Featuring is a great way to highlight and showcase outstanding student work throughout the week and can be done on a unlimited number of student posts. Showcasing one or multiple students posts throughout the week can help give students within the community an understanding of the quality of work that should be going into their post!
What do our Algorithms look for?
- The length and depth of the post
- The repetitiveness of the text
- The way the student communicates in their post (e.g. images, video, clear formatting)
- If the post is flagged due to Community Guidelines, or was similar to other posts that did not meet Community Guidelines
- If the post is similar to other posts that were featured in the past
In Moderation
When Packback’s AI flags a post that has violated Packbacks community guidelines our moderation team will then review the flagged post and decide to moderate it or not based on our Community Guidelines.
This process is considered AI-assisted, since a member of the Packback team still carefully reviews all flagged posts to ensure that no posts are mistakenly moderated. For more information on why Packback flags and moderates post, please check out our article on: Introduction to Packback’s Auto-Flagging & Moderation
How AI-feedback is used to support students on Packback
Packbacks instant feedback is designed to support Mastery Learning, our goal is to provide non-punitive feedback that encourages growth over feedback solely for assessment purposes.
This connects very closely to the work of Bloom’s 2 Sigma Problem. Bloom found that students perform 2 standard deviations better when they receive individualized feedback and chances to retry their work! Bloom knew at the time of his experimentation that the kind of 1:1 feedback needed to achieve the results demonstrated in the 2 sigma problem becomes unmanageable for instructors teaching more than a few students. He knew that one day, technology would improve to a point where individualized, instant feedback with multiple chances for the students to retry and improve would be possible without putting an untenable burden on instructors.
Packback employs a pedagogical approach that aligns with the Community of Inquiry model, wherein students gain value not just from participating in the discussion and writing their own responses, but from self-regulating and guiding the discussion through the process of asking questions.
AI based feedback allows every student to receive the same amount of individualized, mastery-focused feedback when engaging in the discussion, no matter how large the course size!
How AI is used on Packback to support Instructors
AI presents an incredible opportunity to allow instructors to choose assignments based on their pedagogical merit, not based on the limitations of their time for managing and grading; particularly in larger courses.
On Packback our AI relieves the manually labor of having to spot check every students post for quality of work by reviewing your students posts for you! After our AI reviews a students post, that student will then receive a coaching email to why their posts has been moderated after being reviewed by our AI. Our AI is your quality enforcement!
A huge body of research supports the pedagogical merits of student-led asynchronous discussion, but this is often one of the first assignments to be removed when course sizes increase above 30 or 40 students because of the sheer volume of work generated in this style of discussion that historically instructors would need to review one-by-one. Using Packbacks AI to handle some of the more foundational or repetitive tasks of managing a discussion means instructors are able to spend less time grading, and more time guiding the discussion through the use of Custom Feedback, Guiding Questions, Featuring, and selecting posts to bring back into the synchronous course discussion.
Packback’s AI Ethics policy
At Packback we believe AI should be used to augment educators to make their lives better, and allow students to receive instant feedback when and where they need it. When implemented and designed carefully and intentionally, tools like Packback present an opportunity to make great education more accessible for more students than ever before, while making educator’s lives easier and their jobs more enjoyable. But developing any algorithmic technology that has an impact on real peoples’ lives comes with a responsibility to ensure the algorithms are transparent, held accountable by human beings, and considerate of avoiding bias in the training and development of the models. To ensure that our continued work to make pedagogical best practices possible and easier to manage at scale through technology, Packback’s AI ethics policy guides our work. For more information please review our Ethics Policy.