The LLM Critics Are Right. I Use LLMs Anyway

	The LLM Critics Are Right. I Use LLMs Anyway(theocharis.dev)
	239 points by JeremyTheo 23 hours ago \| 246 comments
	tl;dr: The author acknowledges that most criticisms of LLMs (slop, environmental cost, ethics, killing OSS trust, undermining junior engineers) are valid, yet still uses them heavily because they amplify existing thinking rather than replace it. Key patterns that work: forcing the LLM to grill you with questions before acting, using adversarial subagents to tear apart your work, and only using LLMs in domains where you can distinguish good output from slop. The dividing line between "AI slop" and useful output is whether a human put real thought behind it—which is invisible from the outside, making trust everything.
	HN Discussion: ~Concern that heavy LLM use will atrophy engineering skills and thinking over time ↓LLMs are dangerously anthropomorphized and not comparable to past tools like calculators ↑LLMs genuinely make work faster and higher quality, confirming the author's experience ↑LLMs should be embraced as accelerators of learning and human progress ↓The article's specific claims (geopolitics, junior engineers, uniform critic views) are inaccurate or oversimplified