New from me at LeadDev, Your error budgets don't know AI exists.
The argument is not that AI-generated software is inherently less reliable. That frame is too easy.
The harder problem is that AI has separated the cost of producing change from the cost of understanding it. A team can ship more often, recover quickly, keep its SLO intact, and still lack the mental models required to explain consequential behavior when something unexpected happens.
Error budgets still price failure. They do not tell us whether the organization still understands the system it is accountable for.
Read the full piece at LeadDev. The full text will run here after LeadDev's exclusivity window closes in late November.