# Sam Altman: AI Will Replace Many Support Jobs, Upend Coding Careers, Spare Nurses > AI will automate Tier-1 support-passwords, refunds, lookups-Altman says. Humans move to empathy, judgment, complex cases, and supervising AI quality and handoffs. - Type: AI news - Page: https://completeaitraining.com/news/sam-altman-ai-will-replace-many-support-jobs-upend-coding/ - Published: 2025-09-13 - Updated: 2025-09-14 - Author: Jeroen Erne - Jobs: Customer support - Reading time: 3 min ### AI And Customer Support: What Sam Altman Just Signaled OpenAI CEO Sam Altman recently said the quiet part out loud: "I'm confident that a lot of current customer support… will be better done by an AI." He also noted that no one can predict the future perfectly, but he's clear that first-line support is in the crosshairs. He drew a contrast with roles that rely on deep human connection, using nursing as the example. Demis Hassabis from DeepMind echoed the same: empathy-led care is different. For support pros, the message is simple-routine tasks get automated; human trust, judgment, and complexity remain. #### What This Means For Your Job Tier-1 tickets and repetitive workflows will be automated. Human roles shift toward higher-value work, supervision, and complex cases. - AI will handle password resets, order status, shipping ETA, refunds/returns, and policy lookups. - AI will draft replies, summarize threads, translate messages, and classify intents. - Humans will own de-escalation, exception approvals, multi-system troubleshooting, and account retention. #### Where Humans Win Altman and Hassabis both point to the same advantage: empathy and trust. That's your moat-especially in moments that actually change outcomes. - High-emotion conversations, churn saves, and sensitive issues. - Ambiguous problems with messy context or multiple tools involved. - Policy judgment, edge cases, and risk calls. - Relationship work with key customers and complex B2B accounts. #### Shift From "Answering" To "Orchestrating" The durable role is the AI-enabled support specialist. You design workflows, supervise AI, and ensure quality and compliance. - Conversation design and guardrails: define tone, boundaries, and handoff triggers. - Knowledge ops: keep articles structured, up-to-date, and optimized for AI retrieval. - Human-in-the-loop QA: review AI responses, fix gaps, and train better prompts. - Escalation strategy: decide what never gets automated and why. #### 30-60-90 Day Plan - **Days 0-30:** Audit top 20 intents and macros. Baseline CSAT, FCR, AHT, deflection, and QA scores. Use an AI assistant to draft replies and summaries; keep a prompt library. Identify 5 policy risks and write clear guidance. - **Days 31-60:** Pilot AI on one low-risk queue with human review. Set handoff rules (emotion detected, policy mentions, VIP flag). Compare pilot KPIs to baseline. Improve knowledge articles for clarity and structure. - **Days 61-90:** Automate 25-35% of Tier-1. Formalize weekly QA of AI outputs. Publish an escalation playbook. Propose a new scope for your role: AI supervisor + complex case lead. #### Tools To Test Start with a general AI assistant for drafting and classification and your help desk's native AI features. Keep customer data safe: anonymize PII and follow company policy. Explore [ChatGPT](https://openai.com/chatgpt?ref=completeaitraining.com) for summaries, translations, tone adjustments, and intent tagging. Pair it with your ticketing system's automation (triggers, SLAs, macros) for fast wins. #### Prompt Patterns That Work In Support - **Classification:** "Classify this ticket into one intent from \[list\]. Output JSON with intent and confidence." - **Troubleshooting:** "Generate a step-by-step checklist. Ask one question at a time. Stop and wait for user reply." - **Empathy-first reply:** "Write a 3-sentence response: 1) acknowledge emotion, 2) state what you'll do, 3) set next step. Keep it under 80 words." - **Policy-safe response:** "Answer only using this policy summary. If uncertain, ask for escalation." #### Metrics That Matter With AI - First Contact Resolution (FCR) and deflection rate (without hurting CSAT) - CSAT and QA score for both AI-drafted and human-edited replies - Average Handle Time (AHT) for complex vs. simple tickets - Escalation accuracy (right cases escalated at the right moment) #### Skills To Build This Year - Prompt writing and conversation design - Knowledge base structuring and maintenance - De-escalation, empathy, and negotiation - Policy, compliance, and risk awareness - Data literacy: read dashboards, run small experiments, report impact #### Programmers And Nurses: The Contrast Altman said he's less certain about the future shape of programming jobs as AI makes software creation easier. Demand could keep roles strong-or automation could compress them over time. For nurses, he expects minimal change because people want "the deep human connection." That's your cue: make empathy and judgment the center of your work, then use AI to clear the busywork. #### Practical Next Steps - Pick one queue and target 30% automation with human review. - Standardize prompts and macros; log what works and what fails. - Codify escalation criteria to protect customers and compliance. - Report weekly impact to leadership with hard numbers and examples. #### Keep Learning If you want structured training for support workflows, see our curated options: [Courses by Job](https://completeaitraining.com/courses-by-job/) and [ChatGPT resources](https://completeaitraining.com/tag/chatgpt/). #### The Bottom Line Altman's message is clear: AI will take a big bite out of routine support. Your advantage is the human side-empathy, judgment, and ownership of complex outcomes. Become the person who designs, supervises, and improves the system. That's how you stay valuable while the work shifts. --- This is the Markdown form of https://completeaitraining.com/news/sam-altman-ai-will-replace-many-support-jobs-upend-coding/. The whole site is mapped at https://completeaitraining.com/llms.txt.