What we expect from an AI tutor Expert What we expect from an AI tutor Expert

What we expect from an AI tutor Expert

Here is an article that describes the essential role of data annotators in developing AI models, focusing on their day-to-day tasks such as evaluating responses for relevance, coherence, and tone, and their broader impact in ensuring AI systems are accurate, ethical, and effective across various applications.

 

1. Desire to understand the ask

We want you to ask any questions you have about the task. If something is unclear or needs more explanation, feel free to write in the task chat. We support sharing experiences and knowledge among experts annotators. If you see a question in the chat that you can answer, please help.

 

2. Quality

Our aim is always to get good-quality evaluations. We check quality using different methods:

  • Automatic checks.

  • Analysis by the auditors

If we notice that your quality isn't high enough, for example, if you have only 6 correct answers out of 10, the task team will contact you to explain your mistakes. Please pay close attention to this feedback.

In case your quality consistently falls below the acceptable level, we will no longer be able to involve you in our AI tutor Expert Annotator tasks.

 

3. Human content

Since our tasks often involve improving algorithms, including multi-turn learning systems, we expect the content you create in tasks to be a result of your creativity. Avoid using AI bots, like ChatGPT, when doing tasks unless you are specifically instructed to use it.

 

4. Questions and feedback

Your feedback and inquiries enable us to identify and address any weaknesses in our instructions, enhancing our overall performance and facilitating discussions with our team. We highly value your input and encourage you to share your thoughts with us.

 

❗ We do not tolerate any forms of cheating, including but not limited to using tools for automatic labeling, randomly clicking through task answers, or skipping all complex tasks to focus solely on simple ones.

Was this article helpful?

156 out of 164 found this helpful

Articles in this section

Add comment

Article is closed for comments.