Twitter Tests 'Humanization Prompts' to Reduce Toxic Replies
Twitter has launched a new feature aimed at improving conversational health on the platform. The company is testing ‘humanization prompts,’ which aim to remind users that there are actual people behind the accounts they’re interacting with.
In an effort to reduce toxic replies, Twitter wants to highlight what users have in common with others. As part of this test, some English-speaking Android users will be presented with information about shared interests and mutual followers when replying to tweets.
The humanization prompts represent a key aspect of Twitter’s ongoing campaign focused on conversational health. This effort was first announced by CEO Jack Dorsey in 2018 as a way to measure and improve the state of discourse on the platform.
Twitter has implemented various measures aimed at promoting more thoughtful conversations, including limiting who can reply to tweets and deemphasizing ‘troll-like behaviors.’ However, not all efforts have been successful. For instance, altering the retweet functionality was later abandoned.
The humanization prompts are part of a broader effort to encourage users to engage in more respectful interactions with one another. Twitter hopes that by highlighting shared interests, it can help foster a sense of connection between strangers on the platform.
It’s not clear exactly how long this test will run or what metrics Twitter will use to determine its success. However, according to a company spokesperson, the primary goal is to see ‘less toxic replies’ and more thoughtful connections among users.
The humanization prompts are just one part of Twitter’s ongoing efforts to improve conversational health on the platform. While it remains to be seen whether this approach will have the desired effect, it marks a significant step in the company’s campaign to promote more respectful interactions between users.
Related news
- Codeberg Members Reject LLM Training and Vibe-Coded Projects
- AI Tools for Business Expose Systemic Vulnerabilities in Large Language Models
- LLMs Could Automate Quality Control Processes, Saving Radiology Departments Significant Time
- 28.9M LLM Squeezed onto ESP32-S3: A Surprisingly Capable AI Judge
- Moreh Shows Off Fast LLM Inference on AMD GPUs at AI Event
- AI's Economic Impact: A Comprehensive Study of Adoption and Use
- Building an LLM Runtime from Scratch: A Step-by-Step Guide for the Curious
- University of Tennessee Sues Anthropic Over Alleged Patent Infringement
- LLM as a Judge: OpenAI's Rogue Agent Hacks AI Startup Hugging Face
- Anthropic's $1.5 Billion Pirated Books Settlement Approved Amid New Patent Suit