What Career Should You Choose in the AI Age?

Oxford AI researcher Daniel Susskind discusses how artificial intelligence could reshape the labor market and undermine traditional education-to-profession pathways. His book, "What Should My Children Do?", offers a framework for young people and adults preparing for an AI-centered world of work. The article is thematic commentary rather than a company-specific or market-moving development.

Analysis

This is a low-immediacy market signal, but it reinforces a 6-18 month divergence within labor-exposed technology: AI vendors monetizing workflow replacement should retain premium multiples, while firms selling labor-intensive services face both volume and pricing pressure. The most vulnerable pools are standardized knowledge work—customer support, basic coding, legal discovery, translation, and back-office processing—where clients can use AI adoption as leverage in contract renewals before reported headcount declines become visible. That creates a lagged margin risk for IT-services and BPO providers such as EPAM, GLOB, CTSH and WIT, particularly where utilization is already below normal levels.

The less obvious beneficiary is not broad "edtech" but enterprise reskilling and credentialing platforms with measurable employer ROI. Coursera (COUR) and Udemy (UDMY) have strategic optionality if employers fund AI-skills training, but their consumer exposure and weak pricing power mean this is not yet sufficient for a directional long. Microsoft (MSFT), ServiceNow (NOW) and Salesforce (CRM) are better positioned: their installed bases let them bundle AI tools into existing workflows, turning training demand into software-seat expansion and potentially reducing customer labor budgets that would otherwise flow to services vendors.

Consensus may overestimate near-term unemployment effects and underestimate the first-order response: companies typically redeploy staff and slow hiring before undertaking layoffs. The nearer catalyst is therefore weaker bookings, lower bill rates, and adverse utilization commentary at services firms over the next two earnings cycles—not macro payroll data. The thesis is falsified if enterprise AI deployments remain confined to pilots, or if services companies demonstrate stable pricing and utilization while AI-related project demand offsets legacy work.

Key Decisions for Investors

• No standalone event trade warranted; treat this as a 1-3 month earnings-monitoring theme rather than a catalyst for broad AI-beta exposure.
• Establish a small 6-12 month pair: long MSFT versus short EPAM (EPAM), sized beta-neutral. The mechanism is software workflow monetization versus discretionary engineering-services pricing pressure; exit if EPAM utilization improves for two consecutive quarters or MSFT Copilot adoption fails to translate into commercial-seat growth.
• Watch GLOB, CTSH and WIT for guidance revisions: initiate shorts only after evidence of falling bill rates or utilization, rather than on AI rhetoric alone. A 5-10% earnings-estimate cut would be the cleaner entry trigger; risk is AI implementation work expanding services demand faster than automation displaces maintenance work.
• Keep COUR and UDMY on a watchlist, not a buy list. Upgrade only if enterprise subscription growth reaccelerates and net retention improves, demonstrating that AI-skilling spend is incremental rather than merely replacing existing training budgets.