A US AT&T call center worker details working with an AI tool that transcribes calls and offers technical solutions, and wonders “am I training my replacement?”
To many people, chatbots and other technology feel like a ticking time bomb, sure to explode their work.
Context & Ripple Effects
This is the human ledger of a shift that began years ago: pandemic-era budget cuts pushed call centers toward chatbots like IBM's Watson to filter calls and cut operator headcount, and by early 2023 the Wall Street Journal found conversational AI splitting US agents into those it helps and those it leaves with harder, more intense calls. The AT&T worker's account adds the sharpest version of that split — the same tool that assists her is built from her own calls.
What makes the piece land is that her labor now has a dual output: resolved tickets and training data. Two years later, agents across four countries report customers mistaking them for AI, a sign the human role is being reshaped around the tool rather than the other way round.
First-order effects
- For the AT&T agent and her peers, every call now serves two masters: the customer on the line and the transcription-and-suggestions model learning from the transcript, so assisting the tool and competing with it happen in the same shift.
Second-order effects
- Rival carriers and outsourcers watching AT&T's deployment face pressure to adopt similar copilots simply to hold handle-time and cost-per-call parity, while agents left with the residual complex calls — the pattern the WSJ documented — push back on workloads that automation concentrates rather than relieves.
Third-order effects
- If the pattern holds, the call center job description inverts from resolving calls to supervising them: the 2026 reporting on "AI natives" needing careful oversight of their tools points to a structure where human labor is priced as quality control on AI output, and the worker's question — training her replacement — becomes a standing feature of the role rather than a one-time fear.
The trend: Call centers are moving from AI as an agent's assistant to agents as AI's supervisors and training data, with each assisted call tightening that loop.