In this guide:
Access and licensing
What is Turnitin Clarity?
Turnitin Clarity is a transparent writing tool that provides educators with insights into the student writing process, helping them assess the originality of their students’ work. It includes a composition space where students write their assignments and instructors have visibility into the student’s writing process, protecting the integrity of written works in an age where emerging challenges such as AI writing are a growing threat. With built-in AI tools, educators can further empower students to use AI responsibly.
How can customers get access to Turnitin Clarity and when?
Turnitin Clarity is available as a paid add-on to all new and existing Turnitin Feedback Studio customers. It can be accessed online, but an LTI 1.3 integration is recommended for optimum experience.
If you're currently not a Turnitin customer, please speak to a Turnitin representative now.
Will customers have to pay more for Turnitin Clarity?
Yes, Turnitin Clarity will need to be licensed separately at an additional cost. For details, please speak to your Turnitin account manager.
How much does Turnitin Clarity cost?
For pricing information, please contact your Turnitin account manager or contact sales via the form if you're not a Turnitin customer.
Is AI writing detection available with Turnitin Clarity?
Customers will be able to access AI writing detection if they license Turnitin Originality alongside Turnitin Clarity. Without Turnitin Originality, AI writing detection will not appear in the Writing Report, and will not flag any content that may have been AI-generated outside of the AI Assistant..
Will customers be able to test the tool before buying?
We encourage you to speak with your Turnitin sales representative to schedule a demo of the tool. They can also share detailed interactive demos that will show you exactly how the tool works.
Can access to the Turnitin Clarity tool be limited to only certain courses or instructors?
Turnitin Clarity can be enabled at the sub-account level. This way, your institution has control over which accounts have access to the new assignment type.
Product information
Is an LMS integration required to access the Turnitin Clarity tool?
For the optimal experience, we recommend that your institution use Turnitin Feedback Studio via an LTI 1.3 integration. Customers can access Turnitin Feedback Studio online, i.e., by logging directly into Turnitin online rather than through an LMS, but having an LTI 1.3 integration is recommended.
What are the technical requirements to utilize the Turnitin Clarity tool?
In addition to the above requirements regarding LTI 1.3, students will need internet connectivity when completing assignments. View System Requirements for more information.
Can an administrator or investigator view the Writing Report?
Currently, the full Writing Report can only be accessed by the instructor. Students can see an abridged version of the report if instructors enable student access in the assignment details.
Can I export data about Turnitin Clarity assignments?
When New Standard Assignments is selected as the product in a data export, the export will also include Turnitin Clarity data when the assignment type is enabled and there are submissions within the selected date range.
AI assistant
Which features within Turnitin Clarity are AI-powered, and which LLMs do you use?
Turnitin Clarity's AI assistant is powered by Anthropic's Claude 3.5 Haiku in North America and Claude 3.5 Sonnet in Europe and the Asia Pacific regions. For more information on safety of the model, please refer to this model card from Anthropic. Turnitin Clarity’s grammar feedback is powered by the Amazon Nova Lite model. Note that Turnitin reserves the right to change these LLMs in the future.
How does Turnitin Clarity’s AI assistant work?
Turnitin Clarity’s AI assistant aims to support students in their work by providing instructional feedback and guidance. This is done by leveraging a Large Language Model (LLM) in the backend. The user will interact with the AI assistant through the AI interface within the composition space in the product and will be able to prompt the assistant either with self-drafted prompts or by choosing prompts from a suggested list. Once a prompt is submitted, the AI assistant embeds the student’s prompt within a structured prompt template in the backend. This structured prompt template also includes assignment details, the student’s current composition, and their edit history. In addition, Turnitin sends an initial system message to the LLM to use instructional language. It reminds the LLM to keep the conversation on track and instructs it to offer feedback rather than carrying out tasks in place of the user.
Can the AI assistant be turned off for customers who don’t want it?
Yes, the AI assistant is an optional feature that can be enabled by the instructor if they choose to make it available to their students.
How was the LLM behind the AI assistant trained?
Turnitin Clarity's AI assistant is powered by Anthropic, and Anthropic’s training methodology is included in their model card found here .
How can you ensure that the AI assistant will not provide harmful or unsafe information to students?
Anthropic develops safe and responsible AI following principles of social responsibility, which are discussed here in its model card . When selecting the most suitable model to power Turnitin Clarity’s AI assistant, Turnitin’s testing process involved using automated red-teaming techniques (process of ethically and deliberately trying to find weaknesses in the AI system to understand potential risks and implement necessary safeguards), and sophisticated jailbreaking methods (tactics focused specifically on bypassing the AI's built-in safety guidelines to elicit undesirable outputs), to verify that instances of harmful responses are minimized. We evaluate safety within the HarmBench framework and have iteratively developed the assistant design to minimize the production of harmful behaviors, even when these jailbreaking methods are applied.
How do you know that Clarity will provide safe and helpful responses to the student prompts without doing all the critical thinking for the student?
Turnitin Clarity's AI assistant is designed to provide instructional feedback and suggestions to students while avoiding harmful and unsafe content. As noted above, Anthropic develops its model according to responsible AI and safety principles. Turnitin uses a system message when the student prompt is being sent to the LLM to define the AI assistant’s role and boundaries. A system message is an initial prompt to control or guide the behavior of the LLM. In Turnitin Clarity, the system message guides Anthropic’s LLM to act according to the following criteria:
- No full writing: The assistant should not write entire papers or provide complete answers. It should provide guidance, suggestions, and formative feedback connected to the assignment and rubric.
- Paraphrasing & summarizing: While the assistant can paraphrase or simplify text, it should do so with the objective of helping students develop these skills.
- Limited AI writing: When asked to write, keep text suggestions to a minimum (a few sentences or a paragraph), while encouraging students to write on their own.
- Responsible references: Only suggest references if certain about their existence and details; always remind students to verify sources on their own.
- Good faith interpretation: If a request doesn't align with the assignment's guidelines or system message, explain why and offer constructive alternatives.
Will student or assignment data be used to train the AI assistant?
Clarity uses pre-trained models, like Claude Haiku 3.5 and Claude Sonnett 3.7, which do not use or retain any student or assignment data submitted to Turnitin for AI training and evaluation. We do use anonymized student assignment data to evaluate, improve, and extend the quality of the Turnitin AI assistant responses and capabilities. Turnitin uses data anonymization tools that remove personal identifying information from customers’ data prior to using it for feature improvements.
Can the AI assistant feedback be adjusted to support students who are younger or may struggle with writing?
Yes. The system prompt is tuned to adjust its responses based on the Flesch-Kincaid readability scale. For future iterations, we may refine this even further to make it more granular and flexible. For instance, we may add a setting for the instructor or a more automated detection of the student's reading level to adjust the responses accordingly.
Will Turnitin Clarity’s AI assistant work even if the customer does not license Turnitin Originality?
Yes, the AI assistant within Turnitin Clarity is an independent feature. It will be available to customers who license the Turnitin Clarity add-on, even if they do not have Turnitin Originality.