In the evolving landscape of journalism, artificial intelligence (AI) is increasingly becoming a tool for reporters, especially those working independently without the traditional support systems of newsrooms. A notable example is technology reporter Alex Heath, who has integrated AI deeply into his writing process. Since leaving traditional media to work independently on Substack last year, Heath uses Anthropic's AI tool, Claude Cowork, alongside the Wispr Flow voice-to-text service, to draft his articles. Speaking into a microphone, Heath feeds ideas to Claude, which then crafts the first draft of his stories. This AI is connected to various productivity tools like Gmail, Google Calendar, transcription services, and note-taking apps, enabling it to assist comprehensively.
Heath has gone beyond just feeding text to the AI; he has developed a sophisticated "skill" set for Claude that includes detailed instructions on his writing style, previous articles, and guidelines he calls the "10 commandments" of writing like Alex Heath. This customization allows Claude to produce drafts that closely reflect Heath's voice and preferences. After the AI completes a draft, Heath spends up to 30 minutes revising and interacting with the tool-an involved process that still requires his input but significantly reduces the overall writing time.
He estimates that this workflow saves him 30 to 40 percent of the time he used to spend writing, turning what he once disliked-the initial, often daunting stage of drafting-into an enjoyable task. Heath views AI as essential to managing the volume of work required to sustain his independent journalism.
Heath is part of a growing group of tech journalists who are adopting AI to assist in writing and editing, especially those who have lost newsroom resources like editors and fact-checkers. Independent reporters are not simply using AI to generate stories but are creatively reconstructing editorial support systems through AI tools. This shift raises fundamental questions about the role of human journalists. If AI can write, edit, and fact-check, what unique contributions do human reporters make? Research from Google DeepMind warns that over-reliance on AI can lead to writing that is bland, less creative, and more neutral in tone. Journalists assert that to use AI effectively, they must understand the unique value they offer-whether it is exclusive scoops, distinctive voice, or insightful analysis.
For Heath, his primary value lies in uncovering scoops and engaging with sources, rather than prose alone. By delegating drafting to Claude, he can devote more time to reporting and communicating with his audience. Heath's approach recalls the traditional "rewrite desk" from the pre-digital era, when reporters filed stories by phone to editors who crafted the final articles. In this analogy, Claude serves as Heath's modern rewrite desk, allowing him to focus on reporting while the AI handles drafting. Heath admits to feeling like he is "cheating" but finds the process liberating, enabling him to concentrate on what he enjoys most-reporting, learning, and delivering timely, insightful news to readers.
Another journalist experimenting with AI in a different way is Jasmine Sun, who runs a newsletter on AI and Silicon Valley culture. Unlike Heath, Sun does not use AI to write her articles, partly because she believes post-training of AI models can stifle creativity. Instead, she employs Claude exclusively as an editor to help refine her work. Sun feeds Claude her past writings and style notes but instructs the AI explicitly not to write sentences for her or be sycophantic. Instead, Claude's role is to challenge and enhance her voice, providing critical feedback to push her writing toward the best version of herself. Sun emphasizes that Claude is not a co-writer but a rigorous editor designed to elicit her best work through constructive critique.
Sun's use of AI as an editor, rather than a co-writer, has sparked some controversy. Some critics argue that AI cannot replace the transformative role of a human editor who challenges and shapes a writer's ideas deeply. Sun finds this reaction puzzling because many independent writers cannot afford human editors, and Claude's feedback helps maintain high standards. She compares Claude to an advanced version of Grammarly, capable of identifying structural and conceptual weaknesses in writing rather than just grammar and style issues. According to Sun, Claude's willingness to offer blunt feedback makes her writing process more disciplined and rigorous.
Casey Newton, author of the newsletter Platformer, echoes similar sentiments about AI's impact on journalism. He distinguishes between value derived from information and value derived from voice, opinion, and analysis. Newton suggests that if readers primarily seek information, they may be indifferent to AI-generated writing. However, for content that relies on unique voice and insight, AI-generated prose may feel cheap or generic. Newton is not currently using AI to write but has experimented with AI editors modeled after his own writing, inspired by Sun's approach. He finds that AI feedback can rival that of human editors, which has led him to reconsider the balance between original reporting and news analysis in his work, shifting toward more original reporting as AI advances.
Taylor Lorenz, who writes the User Mag Substack, uses AI tools to support the business side of her media work rather than in the writing process itself. She leverages AI to generate SEO-friendly descriptions and sift through data but does not entrust AI with writing or editing articles. Lorenz cites a lack of trust in AI's handling of sensitive reporting materials and a personal love for the craft of writing as reasons for her reluctance. She sees journalism fundamentally as a mission to illuminate the world and does not want AI to replace the human element in that endeavor.
Kevin Roose, a technology columnist for The New York Times, uses AI in his book-writing process to significant effect. He credits AI tools with shortening the time needed to complete his book on the race to develop artificial intelligence by two to three years. Roose has organized a team of Claude agents that perform specialized editing tasks, including fact-checking, ensuring stylistic consistency, and providing both positive and negative feedback. While Roose still collaborates with human editors and writes the book himself, he acknowledges the growing capabilities of AI. For now, he believes his writing surpasses AI's generic and depersonalized output and enjoys the process. Roose anticipates that AI will eventually outperform humans in all aspects of writing but sees the human element-personality, perspective, and connection-as an advantage that remains relevant today.
The experiences shared by these journalists illustrate a spectrum of AI use in journalism, ranging from drafting and editing to business support and feedback mechanisms. While AI tools like Claude are reshaping workflows and challenging traditional notions of writing and editing, journalists emphasize the ongoing importance of human insight, voice, and reporting skills. The integration of AI is not about replacing journalists but augmenting their capabilities, enabling them to focus on areas where human creativity and judgment remain essential.
As AI continues to evolve, independent journalists, in particular, are finding innovative ways to incorporate these technologies to compensate for lost newsroom infrastructure and maintain high-quality output. Whether AI serves as a drafting assistant, a rigorous editor, or a business tool, the key to successful adoption lies in thoughtful use that preserves the unique human elements of journalism. The future of the profession may well depend on the balance between embracing AI's efficiencies and safeguarding the distinct contributions only human journalists can provide.
This overview is based on insights from multiple journalists and reflects ongoing debates about AI's role in the craft and business of journalism.