AI Tools Are 'Deskilling' Workers, Philosophy Professor Says

Anastasia Berg, a UC Irvine philosophy professor, warns that heavy reliance on AI is producing rapid skill attrition among workers—especially junior staff—who may never build foundational abilities to verify or correct AI outputs. Supporting data from a joint analysis of 1.58 million ChatGPT conversations shows 73% of adult messages by June 2025 were non-work related, underscoring pervasive cognitive offloading; the implication for investors is a potential long-term productivity drag and hidden operational risk for firms that equate AI adoption with durable efficiency gains.

Analysis

Market structure: AI infrastructure and governance vendors (NVDA, MSFT, GOOGL, AMZN) are direct winners as firms pay for compute, models and monitoring; staffing, entry-level training and commoditized IT services (RHI, CHGG, UPWK, CTSH) are losers because demand for low-skill human labor can compress. Incumbents with proprietary data + scale gain pricing power in cloud/accelerator markets; smaller service providers face margin pressure and client-side consolidation. On supply/demand, persistent chip/datacenter constraints keep short-term pricing power for Nvidia-class suppliers; medium-term human-capital shortages may reduce aggregate demand intensity for certain services.

Risk assessment: Tail risks include accelerated regulatory action (EU/US rules within 6–18 months), large-scale hallucination litigation, or a productivity shock that lowers corporate hiring and GDP growth by 0.5–1% annually over multiple years. Immediate (days–weeks) moves will track earnings and hiring commentary; short-term (3–12 months) effects show in re-tendering of contracts and training budgets; long-term (1–5 years) risk is structural skill atrophy lowering firm-level operating leverage. Hidden dependencies: data-center power, Nvidia supply, and model-ops tooling; catalysts: high-profile AI failure, major class-action, or regulatory fines.

Trade implications: Tactical longs: allocate to NVDA (infrastructure), MSFT/GOOGL (model + cloud), and PLTR/CRWD for governance and security; tactical shorts: CHGG and UPWK plus select consulting/outsourcing names (CTSH, to a lesser extent) where junior labor is core. Options: favor 9–18 month call LEAPs on NVDA/MSFT (buy 25–35% OTM LEAPs) and 6–12 month put constellations on CHGG/UPWK to cap downside. Rotate into cybersecurity and AI-audit vendors over 3–12 months while paring staffing/edtech exposure.

Contrarian angles: Consensus assumes AI only boosts productivity; markets underprice the new demand for verification, senior talent and reskilling services which could raise wages for seniors and create niche winners (training, model-audit firms) over 12–36 months. Historical parallels: calculators/ATMs didn’t eliminate underlying labor but shifted skill premia; here expect reallocation, not blanket obsolescence. The mispricing opportunity: long governance/security + short commoditized staffing is asymmetric if regulation or failures raise verification costs.

Key Decisions for Investors

• Establish a 2.5% portfolio long split: 50% NVDA, 30% MSFT, 20% GOOGL within 2–6 weeks; add on pullback >10% and target 12-month upside of 30–60%; set trailing stop at -20%.
• Initiate 1–1.5% short exposure to CHGG (0.75%) and UPWK (0.75%) via outright short or buying 6–12 month puts if implied vol cheap; use stop-loss at 25% adverse move and target 30–50% downside within 3–9 months as demand for edtech/gig staffing weakens.
• Allocate 1.5% to cybersecurity/model-governance via equal-weight PANW and CRWD using 12-month call spreads (buy 25% OTM call, sell 60% OTM) to limit capital and target 40–80% relative upside if deskilling raises security/verification spend.
• Construct a pair trade: long 1% NVDA vs short 1% CTSH (or RHI if preferred) to capture secular shift from labor-heavy services to automated infra; rebalance quarterly and close or flip if spread narrows by 50% or on regulatory shock within 6–12 months.