Error budget
An explicit, pre-agreed tolerance for a measurable, typically low-severity risk signal (such as a guardrail near-miss trigger rate) over a given period. Staying within the budget earns a team latitude to move at normal pace; crossing it triggers automatic, pre-agreed tightening of review and deployment pace, replacing an implicit, personal argument about risk tolerance with an explicit, shared, evidence-based one.
Defined in 3 GAGE programs, which carry 4 distinct definitions of it. The wording above is taught in Engineering Judgment and Professional Formation.
How each discipline defines it
The same term does different work depending on who is using it. These are the definitions as each program teaches them, unedited.
An explicit, pre-agreed tolerance for a measurable, typically low-severity risk signal (such as a guardrail near-miss trigger rate) over a given period. Staying within the budget earns a team latitude to move at normal pace; crossing it triggers automatic, pre-agreed tightening of review and deployment pace, replacing an implicit, personal argument about risk tolerance with an explicit, shared, evidence-based one.
A quantified, standing agreement, documented in Google's Site Reliability Engineering practice, that expresses how much unreliability a service can tolerate before feature work pauses in favor of reliability work, functioning as the same kind of pre-agreed, trusted trade this topic's three sizes create for a compliance control.
A design-time accounting of the expected error contribution from each component or stage in a system. Components are compared to identify the largest contributor and to verify that the total meets the application's accuracy requirement.
The amount of failure an SLO permits (the inverse of the target). It functions as permission to ship while healthy and a rule to stop and fix when exhausted, ending the features-versus-reliability argument.
Where it is taught
The exact lessons this term appears in. The first module of every program is free with a free account.
- AI System Reliability and Safety Engineering · Risk, Resilience and Frontier AI, Certified AI Transformation Professional (CATP)
- Getting Engineering to Say Yes to a Control · Change Leadership and Politics, Certified AI Transformation Professional (CATP)
- Where latency and error enter the loop and compound · What a Robot Actually Is, Foundations of Robotics and Physical AI
- Engineering responsibility without slowing the build · Responsible Physical AI in Practice, Engineering Judgment and Professional Formation
Where it shows up at work
A definition is where Error budget starts. The lessons above teach it inside the competencies below, which are what job descriptions ask for and what the GAGE credential records. Every link here is drawn from the same careers map that builds a role's readiness check.
Interviews, facilitates, challenges and secures action across legal, security, product and business teams without owning every decision.
2 lessons on this page teach it as a primary outcome.
Plans for a vendor failure, an unsafe model change or a suspended service: rollback, manual fallback, data export and safe decommissioning.
One lesson on this page teach it as a primary outcome.
Designs tests for factuality, robustness, fairness, safety and abuse resistance with rubrics, baselines and thresholds, and says what a score misses.
One lesson on this page teach it as a primary outcome.
Roles that hire for it
Each role page shows the full competency set, what the seat decides, and a readiness check against your own evidence.
- AI Incident Response Leadneeds it at depth · 1 open posting this week
- AI Vendor Risk Managerneeds working knowledge
- Third-Party Cyber Risk Managerneeds working knowledge
- VP of AI Governance and Ethicsneeds working knowledge
- Responsible AI Leadneeds it at depth · 7 open postings this week
- AI Product Manager, responsible product editionneeds it at depth · 3 open postings this week
How job descriptions say it
Postings rarely use the textbook word. These are the phrases the careers map reads as asking for the competencies above.
Where the frameworks place it: DORA; NIS2; NIST AI RMF MANAGE 4; NIST AI RMF MEASURE 1 and 2.
One interview question about it
Legal, engineering and the business want three different things from one AI project. How do you get to a decision everyone will keep?
Asked of candidates for cross-functional facilitation and influence. Answering it in your own words is exactly how a GAGE topic is passed, and the credential records the pass where an employer can check it.
If you come from another field
Some backgrounds already carry this competency, and the readiness check gives credit for it.
- Compliance and regulatory affairs carries it as a claim to back up: Sitting between legal, business and IT is the daily job.
- Nonprofit program and grants oversight carries it as a claim to back up: Translating between program staff, legal and IT is the position an AI governance lead occupies.
- Project and program management carries it as a claim to back up: Cross-functional facilitation is the job.
Terms it appears with
Not an alphabetical neighbourhood: these are the terms taught in the same lessons, ranked by how often they appear together.