SRC07 — Microsoft Overreliance Review — Scorecard¶
Source¶
| Field | Value |
|---|---|
| Title | Overreliance on AI: Literature Review |
| Publisher | Microsoft Research |
| Authors | Microsoft Aether team |
| Date | June 2022 |
| URL | https://www.microsoft.com/en-us/research/wp-content/uploads/2022/06/Aether-Overreliance-on-AI-Review-Final-6.21.22.pdf |
| Type | Literature review / research paper |
Summary Ratings¶
| Dimension | Rating |
|---|---|
| Reliability | High |
| Relevance | High |
| Missing data | Low — comprehensive review of ~60 papers |
| Measurement bias | Low |
| Selective reporting | Low |
| Randomization | N/A |
| Protocol deviation | N/A |
| COI/Funding | Medium — Microsoft sells AI products |
Rationale¶
| Dimension | Rationale |
|---|---|
| Reliability | Systematic review of ~60 papers; rigorous methodology |
| Relevance | Directly addresses automation bias/overreliance — a core equivalent term for sycophancy effects |
| Bias | Microsoft has commercial interest but research arm maintains academic standards |
Evidence Extracts¶
| Evidence | Summary |
|---|---|
| SRC07-E01 | Overreliance occurs when users accept incorrect AI outputs; recommends awareness training, meta-cognition nudges, and post-deployment monitoring |