Grammarly Abandons AI to Clone Experts Without Their Consent

Grammarly Ends Controversial Use of AI

Grammarly recently announced that it would cease the use of artificial intelligence to clone experts without their prior consent. This decision marks a turning point in how the company intends to manage its Expert Review feature. From now on, Grammarly is committed to redefining this feature by allowing experts to decide whether or not they wish to participate in future AI-related projects.

Suspension of the Expert Review Feature

Grammarly's Expert Review feature, which was supposed to draw inspiration from the contributions of real writers, has been disabled by Superhuman. This feature claimed to offer editing suggestions based on the work of influential figures, such as the editor-in-chief of The Verge and other editorial team members.

Reactions and Apologies from Superhuman

Ailian Gan, who serves as the product management director at Superhuman, stated in a press release that the company had decided to disable the Expert Review feature. This decision follows extensive reflection on how to make this feature more beneficial for users while respecting experts' choices regarding their representation. Gan acknowledged that the company had not met its initial goal and apologized, promising to approach things differently in the future.

Reactions from the CEO of Superhuman

Yesterday, Superhuman introduced an email inbox allowing writers to opt out of the Expert Review. However, the company now admits that this measure was insufficient. Superhuman CEO Shishir Mehrotra also apologized and shared his thoughts on the company's future plans in a post on LinkedIn. He expressed his desire to build a future where experts have the option to choose their participation, shape how their knowledge is used, and control their business model.

Criticism and the Future of the Grammarly Agent

Shishir Mehrotra explained that in August, Superhuman launched a Grammarly agent called Expert Review. This agent relied on publicly available information from third-party large language models (LLMs) to provide writing suggestions inspired by the work of influential voices. However, valid criticisms were raised by experts over the past week, concerned about the misrepresentation of their voices by this agent. Mehrotra emphasized that such critical feedback is essential for improving products and stated that the company takes these concerns very seriously. He acknowledged that the company had failed in terms of representing the voices of experts.

A Vision for the Future of AI at Grammarly

The agent was intended to help users discover relevant influential perspectives and research for their work while providing experts with meaningful ways to build deeper relationships with their fans. Grammarly remains committed to solving the "last mile of AI" by integrating artificial intelligence directly into users' work environments. For millions of people, Grammarly is a reliable writing tool, always ready to assist in every application. The company envisions opening its platform to allow anyone to create agents that function like Grammarly, transforming a simple companion into a full team. Imagine a teacher refining your essay, a sales manager reshaping a client proposal, or a leading expert elevating your work. For experts, this is an opportunity to establish a pervasive connection with users while maintaining control over their participation and business model. Mehrotra expressed excitement about building this future with experts eager to collaborate.