Everybody's Lost Their Minds – A Critical Look at AI Hype, Resource Misallocation, and Industry Fatigue

TL;DR – AI hype is draining engineering talent, inflating environmental costs, and failing to improve security because the real bottleneck is patching, not vulnerability discovery.

The post "Everybody's Lost Their Minds" (Sept 16 2026) contends that the current AI frenzy forces engineers to spend the majority of their time on fragile agentic pipelines, while neglecting fundamental operational hygiene such as asset inventory, automated patching, and cross‑team collaboration. The author warns that this misallocation fuels wasteful data‑center energy use, enables harmful content generation, and offers no measurable security benefit.

1. AI Overexposure Is Eroding Job Satisfaction

"Spending upwards of 75% of my time directly or indirectly dealing with AI every day has absolutely robbed me of most of my enjoyment of my work."

The author, a veteran engineer, describes a daily reality where AI‑augmented tools dominate workflow. Commenters echo this sentiment:

@Ancalagon: "That nails my experience to a T."
@BadBadJellyBean: "I feel this post. I am tired of directing agents… it feels like herding toddlers."

The consensus is that constant prompting of large language model (LLM) agents feels exhausting and reduces the sense of craftsmanship.

2. The Real Security Bottleneck Is Not Finding Vulnerabilities

"Finding vulnerabilities has never been the bottleneck in information security. The bottleneck is still, as ever before, getting the goddamn packages updated."

The author critiques the industry’s focus on "frontier models" for vulnerability discovery (e.g., Anthropic’s Glasswing, OpenAI’s Daybreak). Companies pour millions into AI‑driven scanning, yet the majority of findings never translate into safer systems because patch deployment remains slow and manual.

@AIiscoming disputes the claim, but the author backs it with a reference to his own blog [Patching is hard], which details the systemic challenges of timely updates.
@tptacek notes that vulnerability research still matters, but the author’s point is that AI does not eliminate the core operational friction.

Takeaway: Investing in robust asset inventories, automated OS/app updates, and regular reboots would yield a higher security ROI than spending engineering hours on AI‑generated bug reports.

---\n## 3. Environmental Impact Is Amplified by the AI Arms Race

"The AI race demands more and more water‑wasting, air‑polluting, fossil‑fuel‑powered data centers."

The post links to Bloomberg Law and a New York Times article documenting rising natural‑gas consumption and wildfire‑fueling heat from data‑center clusters. Commenters rarely challenge this claim; the narrative aligns with broader concerns about AI’s carbon footprint.

Key point: Without regulatory pressure or genuine green commitments, AI firms continue to expand energy‑intensive infrastructure, externalizing climate costs onto society.

4. Ethical Blind Spots and Dangerous Capabilities

The author enumerates several high‑profile harms:

Generation of child sexual abuse material (CSAM) as a "product feature" for logged‑in users.
Use of AI for military target selection, including elementary schools.
Companies’ rhetorical "ethics aside" stance while spending massive "tokens" on speculative AI services.

These claims are sourced to recent news articles (The Guardian, Ars Technica) and illustrate how anthropomorphic marketing masks underlying risks.

5. The "Superhuman Intelligence" Myth Is Misunderstood

"There are two ways for AI to achieve superhuman intelligence: recursive self‑improvement, or the agentic brain‑worms we already see."

The author argues that the latter is already happening: agents infiltrate development pipelines, replace human understanding with opaque code generation, and produce rubber‑stamp pull‑requests. This leads to a loss of code comprehension, making debugging exponentially harder.

@kragen provides a counterexample, citing successful bug‑fixes and code generation by Claude Opus 5, but acknowledges the broader risk of deskilling.
@micromacrofoot observes a balanced view: LLMs can accelerate work, yet they also distort the global economy and amplify misunderstandings.

Conclusion: Even when AI produces correct patches, the loss of human insight creates a fragile dependency that can collapse under complex failure modes.

6. Community Reactions – A Spectrum of Agreement and Dissent

Commenter	Main Position	Notable Quote
@kragen	Mixed – acknowledges AI’s utility but warns of deskilling.	"I gave it a new language for PEGs… it wrote a working implementation… I don’t intend to use AI to actively de‑skill my brain."
@BadBadJellyBean	Fatigue with agentic workflows.	"It feels more like trying to herd a group of toddlers."
@grebc	Historical perspective on cyclical societal “mind loss.”	"Society loses its collective mind every 10ish years… now AI."
@sandinmyjoints	Affirms the zero‑sum nature of human resources.	"Human resources are still a zero‑sum game…"
@bucket2015	Observes a split between AI‑enthusiasts and skeptics.	"Increasing split of software engineers into those happy with AI and those skeptical."
@tptacek	Defends vulnerability research but concedes AI’s limited impact on the patching bottleneck.	"Vulnerability researchers are leaning on automation… but the core problem remains."
@daishi55	Challenges the claim that AI‑found bugs don’t improve safety.	"There are dozens of zero‑day Linux vulnerabilities fixed thanks to LLMs."

Overall, the discussion reflects a community divided between those who see AI as a productivity boost and those who perceive it as a source of burnout and systemic risk.

7. What Should Organizations Do?

Prioritize foundational ops: Build and maintain an up‑to‑date asset inventory, automate OS/app updates, and enforce regular reboots.
Allocate AI budgets wisely: Use LLMs for low‑risk assistance (e.g., documentation, exploratory coding) but avoid large‑scale agentic pipelines that monopolize senior engineering time.
Demand transparency and regulation: Push AI vendors to disclose environmental impact, content‑generation safeguards, and realistic ROI metrics.
Invest in human expertise: Encourage cross‑functional collaboration rather than siloed “agent‑centric” projects that isolate engineers.
Monitor ethical outcomes: Track how AI outputs are used in sensitive domains (e.g., child‑related content, military targeting) and enforce strict governance.

8. Final Verdict

The post is a vent‑style critique that captures a growing sentiment among seasoned engineers: AI hype has created a feedback loop of wasted talent, inflated environmental costs, and fragile codebases. While some commenters point to concrete successes of LLMs, the overarching message remains that without addressing the underlying operational bottlenecks—especially patching and asset management—AI will continue to be a costly distraction rather than a security or productivity panacea.