When AI is making decisions about people.
Hiring, termination, performance reviews, content moderation, customer support for users in distress, trust and safety review of harm reports, drafted feedback and communications. Contexts where AI's structural limitations meet stakes that demand human judgment.
AI is increasingly inserted into decisions that affect people.
Microsoft Research and Carnegie Mellon (2025), in a peer-reviewed study of 319 knowledge workers across 936 real-world generative AI use cases (Proceedings of CHI '25), found that higher confidence in AI predicted less critical thinking. The authors named this cognitive musculature atrophy.
A 2026 Stanford study published in Science tested 11 leading AI models across nearly 12,000 social situations and found that AI affirms users' actions roughly 50 percent more often than humans do, including in cases involving manipulation or harm. In a follow-up live study, participants who used sycophantic AI came out more convinced they were right and measurably less willing to repair the relationship. They also rated the AI as higher quality and more trustworthy. They could not detect the problem.
Judgment is calibration. When AI handles the routine work, the calibration disappears. Leaders judge exceptions without the base that taught them what normal looks like.
A 60 to 90 minute audit and protocol-building session for senior leaders, grounded in five documented AI failure modes observed during Sāntis Group's development of an AI coaching system. Leaders audit recent AI-mediated decisions, identify which modes are most active, and commit to three judgment protocols for the following 30 days. Not prompt engineering. Intellectual habits and decision standards.
The Completion Problem
AI validates the easy answer instead of interrupting it. A leader frames a performance issue as "this person isn't being a team player." AI completes the pattern within that framing.
The Attribution Problem
AI accepts what the user said about a third party as established fact. A founder describes a CTO who "clearly doesn't want to be here anymore." AI accepts the interpretation as evidence.
The Momentum Problem
AI follows the user's stated direction instead of the direction that matters. The user names what worked and moves past it to the problem.
The Introspection Problem
AI asks for reflection when behavioral evidence is what's needed.
The Confidence Problem
AI produces confident output from degraded input. Missing context and unverified assumptions flow through with no signal.
- Working understanding of all five failure modes with concrete examples from people-impacting decisions
- Audit of recent AI-mediated work against the framework
- Three judgment protocols committed to for 30-day implementation
- Structured intervention against AI-induced judgment erosion across senior leadership
- Early indicator for decision categories at greatest risk: HR, content moderation, customer support, Trust and Safety
- Natural pathway to targeted follow-on coaching
- Standalone workshop. 60 to 90 minutes as part of executive development or senior leadership offsite.
- Extended engagement. 60-minute audit plus 30-day check-in plus 90-day re-assessment.
- Pilot before series. Single-cohort engagement to assess facilitation fit and outcome quality.
- Coaching follow-on. Diagnostic identifies leaders ready for focused 1:1 work; coaching on monthly retainer.
- Series capstone. Final workshop in the Technical EQ series. Series bundle $20,000.
Simone Harvey
PCC-credentialed executive coach (ICF, Georgetown) with 12+ years coaching executives, founders, and senior leaders through high-stakes transitions. At Airbnb and Wells Fargo, designed and ran leadership development programs across thousands of people, including the Black@ Emerging Leaders Sponsorship Program at Airbnb (five 18-month cohorts, 60 percent promotion rate). Twice sponsored for Airbnb's Critical Talent Award.
The five failure modes in this workshop come directly from her development of an AI coaching system, where she observed and field-tested how AI degrades leadership reasoning in decisions about people at scale.
Tell us what you're navigating.
Every enterprise engagement starts with a scoping conversation. We'll respond within two business days with a recommended next step.