Why Temperature Matters? The Three-Temperature Framework: A Pedagogical Model for Controlling Variability in Generative AI
Bibliographic record
Abstract
The behavior of Generative AI models is fundamentally governed by the technical parameter known as temperature, which controls the randomness (variability) of token sampling.1 Because this parameter is poorly understood, users perceive AI outputs as inconsistent, unpredictable, and prone to hallucination, severely undermining trust and complicating regulatory compliance. This concept paper introduces The Three-Temperature Framework, a pedagogical model that translates the abstract technical setting into three concrete, intuitive, and platform-agnostic behavioral modes: Stability Mode (Low Temperature, e.g., $\approx$0–0.3): For tasks demanding precision, factual accuracy, and auditability (e.g., coding, governance-aligned documents). Balanced Mode (Medium Temperature, e.g., $\approx$0.4–0.7): For clear, nuanced, explanatory materials (e.g., teaching, communications). Creativity Mode (High Temperature, e.g., $\approx$0.8–1.0+): For tasks requiring imagination and unconventional ideation. Drawing an analogy from the evolution of early radio technology—where user-friendly controls transformed unstable broadcasts into reliable communication—the Framework provides a conceptual tuning mechanism for managing AI behavior. It directly addresses mandates for transparency, predictability, and robustness found in contemporary regulatory instruments (e.g., the EU AI Act). This paper serves as the foundational document for a proposed multi-site, international study. The Framework was conceived as a direct feedback from students following the author's keynote presentation on AI Governance at a recent multi-school international Hackathon, involving Humber Polytechnic (Canada), CBS International Business School (Germany), University of Salford (U.K.), and GEA College (Slovenia). The author is now seeking broader collaborators to test the Framework across various platforms and institutional contexts.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.008 | 0.017 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.002 | 0.017 |
| Scholarly communication | 0.007 | 0.011 |
| Open science | 0.003 | 0.005 |
| Research integrity | 0.003 | 0.006 |
| Insufficient payload (model declined to judge) | 0.005 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".