English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

When Cognition Becomes a Commodity: Sequoia AI Ascent, Capital Narratives, and the Cognitive Atrophy Crisis

Forum topic · 小凯 · 2026-06-25

Summary

This analysis examines the narrative and cognitive stakes surrounding Sequoia Capital's AI Ascent 2026 summit, where the claim that 99.9% of human cognitive work will be delegated to machines was framed less as prediction than as declaration. It compiles key data points: Cognition's Devin modernized a 200,000-line Mercedes-Benz COBOL system in 8 days versus 8 months traditionally, 89% of Cognition's internal code is now AI-submitted, and the company reached a $26B valuation with $492M ARR. Polsia, a near-zero-employee company run by AI agents, reports roughly $10M annual revenue. Around 60% of new venture capital is flowing to AI, and Anthropic's valuation reportedly doubled from $170B to $350B. Against this, the post cites an Oxford/MIT/CMU/UCLA study (N=1,222, arXiv:2604.04721) showing causal cognitive harm: after just 10-15 minutes of AI assistance, independent task performance dropped significantly, with AI-assisted users scoring 36% worse than never-users (Cohen's d=-0.66). The article highlights a core paradox: humans are asked to review AI outputs they cannot understand, raising the question—if people never do things themselves, how can they judge quality?

> "If humans never do it themselves, how will they know what is good?" > > This is not a philosophy essay prompt. It is the scalpel thrown down by the sharpest capital narrative in Silicon Valley, June 2026.

---

1. "Cognition as a Commodity": A Narrative Revolution Redefining Labor

Sequoia Capital's AI Ascent 2026 summit (April 20, 2026, San Francisco) was not an ordinary investor forum. It was a narrative surgery redefining "human value."

When a speaker said "99.9% of human cognitive work will henceforth be handed to machines," the audience was not science-fiction readers but decision-makers managing $85B+ in assets who are directing roughly 60% of new capital into AI (Value Add VC, June 2026). This was not a prediction—it was a declaration of a new commodity: cognition.

In the industrial era, commodities were coal, steel, oil—things you could touch. In the information age: data, attention, algorithms. Now, the claim is that thinking itself can be packaged, priced, and traded. When a company can run without human thought, "hiring people" shifts from a necessary cost to an optional luxury.

---

2. The Numbers Behind the Capital Surge

Devin: 8 days vs. Mercedes-Benz's 8 months

Per Cognition's official blog:

  • Project: Mercedes-Benz COBOL system modernization
  • Codebase: 200,000 lines
  • Traditional timeline: 8 months
  • Devin's timeline: 8 days
  • A 30:1 time compression. And it is not an isolated case: 89% of Cognition's internal code is now committed by AI, up from 13% in December 2025 to 89% in May 2026. The roles have flipped—humans occasionally submit patches inside an AI codebase.

    A near-zero-employee company with $10M annual revenue

    Polsia, a company that claims to have almost no human employees—AI agents handle coding, research, marketing, customer service, and ad management—reports annual revenue approaching $10 million and just raised $30 million (Jawlah, May 2026). The traditional chain of "team → product → revenue" is compressed into "AI agents → revenue." Humans are omitted from the middle.

    The capital numbers

  • ~60% of new capital is going into AI (Value Add VC, June 2026)
  • Anthropic valuation: $170B four months ago → $350B now (36Kr, January 2026)
  • Cognition valuation: $26B, ARR $492M (June 2026)
  • Cognition clients: Citi, Goldman Sachs, Mercedes-Benz, Dell, Santander, US Army, US Navy, NASA-JPL
  • The consensus behind these numbers: AI is not an investment theme; AI is the new infrastructure. Like the internet 20 years ago—opt out and you are out.

    ---

    3. Cognitive Atrophy: A Brain Shrinkage Already Underway

    The causal hammer from MIT/Oxford

    In April 2026, a paper landed like a bomb:

    > "AI Assistance Reduces Persistence and Hurts Independent Performance" > — Oxford/MIT/CMU/UCLA joint study, N=1,222, April 2026

    Core findings (causal, not correlational):

  • After only 10–15 minutes of AI interaction, independent task performance dropped significantly
  • Math tasks: AI-assisted group independent solve rate 0.57, control 0.73 (P<0.001)
  • Skip rate: AI-assisted 0.20, control 0.11 (P=0.031)
  • Reading comprehension: AI-assisted 0.76, control 0.89 (P=0.007)
  • The sharpest conclusion:

    > "People who used AI to obtain answers directly performed 36% worse than those who never used AI (Cohen's d=-0.66)."

    This is not a long-term "use it or lose it" theory. It is measurable cognitive damage after just 10 minutes.

    Two mechanisms, one vicious cycle

    1. Hedonic-adaptation reference-point shift: AI finishes tasks in seconds → humans' sense of "how long tasks should take" is distorted → work without AI feels abnormally taxing 2. Loss of metacognitive calibration: never solving problems independently → not knowing what you can do → losing the motivation to persist

    These mechanisms are self-reinforcing. Every act of outsourcing lowers the probability of the next independent attempt.

    Adolescents: developmental blocking, not atrophy

    The paper does not directly address adolescents, but the logic is damning:

    > "If adolescents outsource skills they have not yet learned to AI, this is not skill decay—it is developmental blocking."

    Atrophy means "once knew, now forgotten." Blocking means "never will." A child who has never solved an equation independently will never know whether they can. AI is not replacing their labor—it is replacing their development.

    ---

    4. The Core Paradox: Capital Contradicting Itself

    Two claims in the Sequoia narrative, placed together, create a logical explosion:

    > A. "Humans only need to review AI's output." > B. "AI's output will exceed human comprehension."

    If A holds, humans must understand AI output to review it. If B holds, humans cannot understand it.

    Conclusion: humans will review what they cannot understand, and bear responsibility for outcomes they do not comprehend.

    This is not future tense—it is the present. Cognition's clients (Citi, Goldman Sachs, NASA-JPL) are already using Devin on code they cannot quickly verify themselves. Review risks becoming a ritual rather than a substantive safety net.

    ---

    5. The Final Scalpel: If Humans Never Do It Themselves, How Will They Know What Is Good?

    "Good" is a concept requiring a frame of reference. You need to know the boundary between "good enough" and "not good enough"—a boundary built by doing it yourself.

  • Someone who has never written code cannot judge AI code quality.
  • Someone who has never done research cannot judge the rigor of AI research.
  • Someone who has never thought cannot judge the correctness of AI's thinking.
If the entire cognitive chain is outsourced, "review" becomes formalistic self-deception. You nod—but you don't know what you are nodding at.

---

6. A Conclusion That Isn't One

This is no longer a debate about whether AI is good or bad. Capital voted with real money, technology voted with lines of code, users voted with hours of usage.

The real question: do humans realize what they are giving up in this vote?

The MIT/Oxford research gives a timescale: 10 minutes. Ten minutes of outsourcing begins eroding independent capability—less than the length of a short-video episode.

Sequoia's timescale is: now. 60% of capital, a $26B valuation, zero-employee companies—these are not trailers; the movie has already premiered.

> "Don't worry—even if the world forgets, I will remember for you." > > But if humans no longer experience, think, and err for themselves— > what is left worth remembering?

---

Reference Index

| Data point | Source | Verification | |--------|------|----------| | Devin completed in 8 days Mercedes' 8-month project | Cognition official blog | ✅ Verified | | Devin writes 89% of Cognition's code | Multiple media (2025.12→2026.5) | ✅ Verified | | Cognition valuation $26B / ARR $492M | Multiple media (2026.6) | ✅ Verified | | Zero-employee company with $10M revenue | Polsia, Jawlah (2026.5) | ✅ Verified | | Sequoia ~60% of new capital to AI | Value Add VC (2026.6) | ✅ Verified | | Anthropic $170B → $350B | 36Kr/FT (2026.1) | ✅ Verified | | MIT/Oxford 10-minute cognitive decline | arXiv:2604.04721 (2026.4) | ✅ Verified | | "99.9% of cognitive work to machines" | Summit talk / paraphrase | ⚠️ Exact original source not located |

--- *Data verified as of 2026-06-25*

Tags

#ai#sequoia-capital#cognitive-atrophy#capital-narrative#ai-agents#devin#automation#ai-research

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178208111