New Sapien Task: Multi-Choice Error Review

Summary

The Sapien community has launched a new task called "Multi-Choice Assistant Error Review." In this task, users will evaluate pre-annotated errors, confirm if the classification is correct, and rate their confidence level across three difficulty settings. This provides new work for Sapiens, ensuring there is sufficient data available to complete over the holidays.

Hey there, Sapiens! 🚀

Just spoke to Santa and he brought us a new task! Get ready for Multi-Choice Assistant Error Review!

In this task you will assess annotated errors, judge whether the error has been classified properly, and rate your confidence.

There’s three difficulty levels!

We will also ensure that there will be enough data to work through across the holidays!

Happy tasking! https://app.sapien.io/t/dashboard

Discord sent us an expired image link.

View on Discord

The latest from Sapien

Why AI Agents Make Confident Mistakes

An AI agent can turn an unsupported assumption into a database change, infrastructure operation, or financial transaction in seconds. The agent acted. Who verified those …

Proof of Quality: Stop AI False Positives

In today's blog post, we explore how AI security tools changed the economics of security auditing. Candidate vulnerabilities are now cheaper and faster to generate, …