Which entry-level roles AI is eliminating and which remain hard to automate, per 8,000-resume data

Short summary

US tech grads sent 8,000 resumes with almost no response while 90,000 positions were explicitly eliminated due to AI. MIT NANDA data shows 95% of enterprise GenAI deployments return zero P&L, but the 5% that succeed almost all replace repetitive entry-level work like documentation, code review, and customer support. Five roles remain hard to automate: AI trainers, systems integrators, product positioning specialists, and others requiring cross-team judgment and human context.

• •8,000 resumes from top US tech grads met near-silence; 90,000+ positions tagged 'eliminated due to AI'
• •Entry-level docs, code review, support, and analyst roles are being eliminated, not trimmed
• •Five AI-resilient roles require judgment, cross-team communication, and business context that agents can't replicate

Generated with AI, which can make mistakes.