Version 1.0.1: AI Assistant is Back Online
Version 1.0.1 brings back AI Assistant. AI Assistant taps into the power of generative AI to write new exam questions or solutions. This is the tool we used to write most of the solutions and new questions on the platform so far.
AI Assistant Instructions
- Access AI Assistant here or through the navbar under Tools.
- Select an exam => topic => question.
- Select either New Question or New Solution.
- New Question: Use an existing question and solution to write a similar question. Make subsequent edits after the output is generated. The final draft must be approved by an admin before it can be added to the question bank.
- New Solution: Use an existing question and solution to add an alternative solution to an existing question. Once the output is generated, the solution is automatically added as an alternative solution. The solution can be further modified after it has been added.
Notes on Using AI Assistant
Depending on the complexity of the prompt, it can take a few seconds to generate a new question or solution. Results WILL BE LOST if you navigate away from the page. Note that using AI tools consumes tokens. All users begin with 1,000 tokens. Tokens are consumed at the point of entry and cannot be refunded once the request is made.
Please let us know if you need more tokens. We are not stingy with tokens and will happily refill your balance if needed. Since anybody can sign up on the platform for free, the limit is to prevent potential abuse from bad actors running up our data bill.
- To ability to load posts or draft questions from your drafts has been re-enabled.
- The analytics selection dropdown has been reverted back to a series of buttons.
- Tooltips in messenger now show the title of the question a user is working on. This was temporarily disabled.
- Temporarily removed the ability to save GPT model and temperature preferences across instances. This means that if you close the window, your GPT settings will reset to the default setting. The default setting is model: GPT-4 and temperature: 0. We recommend the default setting as it guarantees the most consistent results.
- Other minor UI updates.
- Fixed a bug that was showing the total bookmarks for a question as one less than the actual bookmarks.
- Fixed a bug that was a never-ending loading screen for users who weren't logged in. This applied to the Feed and Mock Exam pages.
- Fixed a bug in the question panel that was distorting the size of the navigation buttons when the question title got too long.
- The desktop navbar menu no longer shows a profile picture. Profile pictures are still shown as part of your profile and can be modified in your settings.