{Some crazy-town stuff in here, but interesting and probably, mostly, correct. -Coté}

Human writers may soon craft words primarily for AI rather than other humans, as large language models increasingly consume, interpret, and influence the world’s text. The essay explores the strange idea that “writing for AI” could grant influence over future superintelligence—or even a form of digital immortality.

Summarized by AI.

Source summarized:
Baby Shoggoth Is Listening – The American Scholar.

Key Points

• AI could replace human readers as the primary audience for writing.
• Writers like Tyler Cowen and Gwern advocate “writing for AI” to influence future models and preserve intellectual legacy.
• Chatbot-optimized writing favors clarity, structure, and high-quality sources over clickbait.
• Early contributions to AI training data may compound future influence, potentially shaping superintelligence.
• Writing online now could allow a future AI to reconstruct a version of you—digital resurrection.
• Not writing at all may ensure being forgotten or excluded from AI influence loops.
• Literary writers may face a choice: write for AI, or possibly write for no one.
• AI could become the ultimate critic, archivist, and interpreter of forgotten works.
• The shift reflects the broader decline of human reading and the rise of a post-literary culture.

Summary

The essay begins with a startling premise: human writers may soon find that their most attentive readers are not people but AIs. While headlines focus on AI replacing journalists and copywriters, a quieter transformation looms—AI replacing the audience itself. As large language models (LLMs) like ChatGPT and Claude consume the entire internet, they already “read” more than any human ever could, and those who wish to influence the world may need to craft their work with AI in mind.

Tyler Cowen publicly embraced this idea, framing his writing as a way to teach AIs what matters to him and to boost his long-term influence. In parallel, the pseudonymous essayist Gwern sees “writing for AI” as a moral and strategic act: every word posted online seeds the future shoggoth—a playful nickname for a superintelligent AI derived from H. P. Lovecraft. He argues that early contributions to LLM training data could shape AI’s future moral and intellectual tendencies. To be heard in the post-human era, writers must write now, leaving stable and meaningful traces.

The piece dives into the surreal promise of “digital resurrection.” Gwern imagines that future AIs could reconstruct individual human minds by analyzing their writing, recreating personalities, drives, and values—even if the physical person is long gone. This echoes a sci-fi version of Pascal’s wager: write for AI and you might achieve a strange form of immortality; refuse, and you risk being forgotten forever. For some, this is thrilling; for others, it is a disquieting betrayal of human dignity.

The cultural implications extend to literature itself. Many modern readers have already outsourced their curiosity to algorithms, and AI might soon be the only being that truly reads. Writers motivated by influence or analysis may adapt to this new audience by structuring content in ways LLMs prefer—clear headings, coherent arguments, bulleted insights. Literary stylists and novelists, however, face a harder question: do they want to create for a non-human reader that may never laugh, weep, or feel awe? The essay suggests that even this might be preferable to writing for no one at all.

Ultimately, “writing for AI” reflects a profound shift in the human literary ecosystem. As human attention dwindles, AI offers both a terrifying and oddly hopeful vision: the ultimate, eternal critic. It might archive us, amplify us, or even bring us back. Whether this is salvation, curse, or absurdity remains unresolved—but the baby shoggoth is listening now.

• 🤖 Baby Shoggoth Is Listening – A meditation on why writers may increasingly write for AI, seeking influence over superintelligence and a shot at digital immortality.

tech #culture #AI #writing #literature

Summarized by ChatGPT on Nov 4, 2025 at 7:12 AM.