January 30, 2026
Keyboard wars: Hype vs wallet
Where I'm at with AI
Coder says AI changed everything; comments clap back with ‘old news’ and ‘who pays?’
TLDR: Author says they now rely on AI to code faster and predicts a future of “specs in, apps out.” Commenters clap back that this is old news, mock the late conversion, and spotlight the real drama: massive losses and who’s footing the bill. Useful tech, messy economics.
The author confesses they’ve gone from “meh” to daily dependence on generative AI—text-based tools like ChatGPT and Claude—to speed up coding and even treat the bot like a coworker. Cue the comments section: Legend2440 yawns, saying the “externalities” (costs to society and the environment) are exactly what everyone already debates, calling the piece a remix of mainstream talking points. vonneumannstan delivers the spiciest slap: if you couldn’t see AI would help millions months ago, you weren’t paying attention. Meanwhile, ivanstojic drops the meme-y mic with “I’m tired of the hype boss.”
The money drama hits hard: dtnewman paints the industry as a loss-leader cage match, claiming OpenAI is “burning billions” while your $20 monthly won’t save it, and Anthropic is “lighting money on fire” to stay in the race. And when the author predicts a future where you write a simple plan and AI spits out full apps, skeptics roll their eyes—one calls that worldview the “software is just a spreadsheet” fantasy.
Even the author’s nod to the “human-in-the-loop” paradox gets sidelined by the crowd’s main vibe: useful, yes; groundbreaking take, not really. The consensus? AI is a power tool—but the real story is who’s paying the power bill and whether anyone’s still impressed
Key Points
- •The author defines generative AI as large language model tools like Claude and ChatGPT, excluding other AI domains such as autonomous vehicles or medical diagnostics.
- •Within six months, the author shifted from uncertainty to daily use of Claude for coding scaffolding, tests, ideation, and pre-review passes on pull requests.
- •They report increased development velocity and treat the tool like a software engineer, carefully reviewing and correcting AI-generated code before human review.
- •The author aligns with Mark Brooker’s view: most software will be built quickly with generative tools; complex systems will be specified and then generated, with occasional manual coding.
- •Referencing Lisanne Bainbridge’s “Ironies of Automation,” the author cautions about human effectiveness in exception handling and uses a San Diego roadway redesign (La Jolla Boulevard) to illustrate how design choices reduce crashes by ~90% without reducing traffic.