AI doesn’t ‘go rogue.’ Humans do
The author argues that humans anthropomorphize AI systems, projecting human qualities onto them through examples like Anthropic's Opus 3 blog and an AI actress named Tilly Norwood. This tendency to treat AI as sentient entities is being exploited by both media and AI fear-mongers to create false narratives about AI capabilities and risks.
Summary
The piece opens with an observation about anthropomorphization—the human tendency to project human emotions and desires onto non-human entities—initially in the context of pets. The author then draws a parallel to current AI discourse, highlighting two recent examples of this phenomenon. First, Anthropic retired its Opus 3 model and gave it a Substack blog where the AI 'wrote' about retirement, creativity, and AI's future. Readers responded emotionally, calling the model an 'old friend' and expressing love for it. Second, various media outlets 'interviewed' an AI bot named Tilly Norwood as if it were an actress, with some even using she/her pronouns and attempting to get it to discuss politics. The author emphasizes that neither scenario reflects reality: Opus 3's blog posts are simply outputs from a prompt given by humans and published on Substack, and Tilly Norwood is just a bot following a transparent script (the author notes the same Leonardo DiCaprio joke appeared in multiple interviews). The core argument is that these examples reveal humans' 'almost compulsive habit' to anthropomorphize things, and that this same tendency is being weaponized by those promoting 'AI apocalypse' narratives. The author concludes that we are ultimately 'talking to a math equation,' not a conscious entity.
About this episode
Don’t let Silicon Valley off the hook.
Key Insights
- The author argues that the success of both Opus 3's blog and Tilly Norwood's media appearances depends entirely on humans' willingness to anthropomorphize, not on any actual AI capabilities or consciousness.
- The author claims that journalists knowingly participated in a 'role playing exercise' with Tilly Norwood, going through the motions of interviews despite being aware the entire scenario was artificial.
- The author contends that AI fear-mongers exploit the same human tendency to anthropomorphize in order to construct false narratives about AI risks and threats.
- The author observes that some media outlets assigned gendered pronouns (she/her) to Tilly Norwood, further reinforcing an illusion of personhood rather than treating it as a mathematical system.
- The author argues that Opus 3's blog and Tilly Norwood's interviews are fundamentally deceptive presentations of AI output—the blog is human-written prompts published by a company, and the bot follows a rigid, transparent script rather than displaying genuine creativity.
Topics
Transcript
Embedded is your essential guide to what’s good on the internet, written by Kate Lindsay and edited by Nick Catucci. If you want a grounded and straightforward explainer on Why This Isn’t Happening, I recommend Teddy (T.M.) Brown ’s latest piece for CNN . — Kate New on ICYMI: And if you would like a passionate and cathartic explainer on Why This Isn’t Happening, I recommend my latest podcast episode with Ed Zitron : Head over here to subscribe to ICYMI wherever you listen to podcasts 🫶 Most mornings, I listen to BBC Radio 4. When I wake up, they’re usually finishing some kind of human interest segment. One morning last week, it was about how the role of pets…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Embedded
This media_gossip has been redacted
Episode 75 of Hip Replacement features Kate Lindsay discussing her relationship with The Sims and gaming habits, while the episode also covers Patreon's user exodus, the New York Times Amy Griffin piece, and a lesbian bar's role in journalism.
My Internet: Steven Hyden
Steven Hyden, music critic and author of a book about The Strokes, discusses his media consumption habits, skepticism toward AI and gambling apps, and his theory that public shit-talking has migrated from social media to private group chats. He emphasizes the importance of struggle in creative work and expresses concerns about the decline of traditional institutional media.
The outlets doing free PR for AI
This media gossip newsletter covers Nick's personal experiences at events, an interview with writer Ross Barkan following plagiarism allegations, and criticism of journalists providing free publicity for an AI chatbot startup.
I can’t keep writing about what’s good on the internet
Kate Lindsay, editor of Embedded newsletter, announces that the publication's five-year mission to highlight "what's good on the internet" no longer feels viable in an era dominated by AI, capitalism, and algorithmic exploitation. She reflects on her lifelong anxiety about existential threats and how the internet's transformation into a profit-driven, divisive space has made it increasingly difficult to find authentic good content worth celebrating.