Language models might be able to self-correct biases—if you ask them
![](https://wp.technologyreview.com/wp-content/uploads/2023/03/self-correct.jpeg)
A study from AI lab Anthropic shows how simple natural-language instructions can steer large language models to produce less toxic content.
![](https://wp.technologyreview.com/wp-content/uploads/2021/08/Casey_web.png)
Articles by Niall Firth
![](https://wp.technologyreview.com/wp-content/uploads/2022/07/Tammy_rgb.png)
Articles by Tammy Xu MIT Technology Review
![](https://neuroleadership.com/wp-content/uploads/2022/08/The-5-Biggest-Biases-That-Affect-Decision-Making.png)
The 5 Biggest Biases That Affect Decision-Making
![](https://lever-client-logos.s3.us-west-2.amazonaws.com/d63643d1-7a20-4e1b-b46d-5308d32d64c2-1622139505411.png)
Anthropic - Research Scientist, Societal Impacts
![](https://wp.technologyreview.com/wp-content/uploads/2020/03/will_heaven-2.png)
Articles by Will Douglas Heaven
![](https://d3535lqr6sqxto.cloudfront.net/employers/X7uBoIliCZjL2SwixXnIYtoUOmsg3Pj962zYnfDD.png)
Research Scientist, Societal Impacts at Anthropic - The Rundown Jobs
![](https://ourworldindata.org/uploads/2022/12/AI-performance_Dynabench-paper-2048x921.png)
language-models/llm-23.md at master · gopala-kr/language-models
![](https://miro.medium.com/v2/resize:fit:1400/1*EUXweZw9pcI_moesu07pRA.png)
My interview with Assistant, a large language model based on ChatGPT-3, developed by OpenAI.
![](https://ars.els-cdn.com/content/image/1-s2.0-S0268401223000233-gr5.jpg)
Opinion Paper: “So what if ChatGPT wrote it?” Multidisciplinary perspectives on opportunities, challenges and implications of generative conversational AI for research, practice and policy - ScienceDirect
![](https://images.prismic.io/encord/760fff56-285d-43d6-8612-d410688a6c71_Vision+Language+Models.png?auto=compress%2Cformat&fit=max)
Unlocking the Power of Vision-Language Models: Understanding Their Mechanisms and Overcoming Challenge
Simon Porter on LinkedIn: Language models might be able to self