Heting Mao
IkanRiddle
·
AI & ML interests
None yet
Recent Activity
reacted
to
kanaria007's
post
with ❤️
about 2 hours ago
✅ New Article: *Designing Ethics Overlays* (v0.1)
Title:
🧩 Designing Ethics Overlays: Constraints, Appeals, and Sandboxes
🔗 https://huggingface.co/blog/kanaria007/designing-ethics-overlay
---
Summary:
“ETH” isn’t a content filter, and it isn’t just prompt hygiene.
This article frames *ethics as runtime governance for effectful actions*: an overlay that can *allow / modify / hard-block / escalate*, while emitting a *traceable EthicsTrace* you can audit and explain.
The key move is to treat safety/rights as *hard constraints or tight ε-bounds*, not a soft “ethics score” that gets traded off against convenience.
> Safety / basic rights are never “weighted-summed” against speed.
> They’re enforced—then you optimize inside the safe set.
---
Why It Matters:
• Prevents silent trade-offs (fairness/privacy/safety “lost in weights”)
• Makes “Why did it say no?” answerable via *machine-grade traces + human-grade explanations*
• Adds *appeals + controlled exceptions (break-glass)* so ETH doesn’t become unchallengeable authority
• Enables safe policy iteration with *ETH sandboxes* (replay/shadow/counterfactual), not blind prod tuning
• Gives operators real KPIs: block rate, appeal outcomes, false positives/negatives, fairness gaps, latency
---
What’s Inside:
• How ETH sits in the runtime loop (OBS → candidates → ETH overlay → RML)
• A layered rule model: *baseline (“never”) / context (“allowed if…”) / grey (“escalate”)*
• Concrete flows: appeal records, exception tokens, SLA-based review loops
• ETH sandbox patterns + an evaluation loop for policy changes
• Performance + failure handling (“hot path”, fail-safe) and common anti-patterns to avoid
---
📖 Structured Intelligence Engineering Series
this is the *how-to-design / how-to-operate* layer for ETH overlays that survive real-world governance.
upvoted
an
article
about 4 hours ago
Designing Ethics Overlays: Constraints, Appeals, and Sandboxes
reacted
to
kanaria007's
post
with ❤️
2 days ago
✅ New Article: *Observations, Under-Observation, and Repair Loops* (v0.1)
Title:
👁️ Observations, Under-Observation, and Repair Loops: The OBS Cookbook for SI-Core
🔗 https://huggingface.co/blog/kanaria007/observations-under-observation
---
Summary:
SI-Core’s rule is simple: *No effectful Jump without PARSED observations.*
This article turns that slogan into an operational design: define *observation units* (sem_type/scope/status/confidence/backing_refs), detect *under-observation* (missing / degraded / biased), and run *repair loops* instead of “jumping in the dark.”
Key clarification: under-observed conditions may still run *read / eval_pre / jump-sandbox*, but must not commit or publish (sandbox: `publish_result=false`, `memory_writes=disabled`).
---
Why It Matters:
• Prevents “we had logs, so we had context” failures: *logs ≠ observations* unless typed + contract-checked
• Makes safety real: even PARSED observations should be gated by *coverage/confidence minima* (declared thresholds)
• Turns OBS into something measurable: *SCover_obs + SInt* become “OBS health” and safe-mode triggers
• Links semantic compression to reality: distinguish *missing raw* vs *compression loss*, and fix the right thing
---
What’s Inside:
• A practical observation-status taxonomy: `PARSED / DEGRADED / STUB / ESTIMATED / MISSING / REDACTED / INVALID` (+ mapping to core status)
• Per-jump *observation contracts* (required sem_types, allowed statuses, age/confidence limits) + explicit fallback actions
• Fallback patterns: *safe-mode / conservative default / sandbox-only / human-in-loop*
• Repair loops as first-class: ledgered `obs.repair_request`, PLB proposals, governance review for contract changes
• Testing OBS itself: property tests, chaos drills, golden-diff for observation streams
---
📖 Structured Intelligence Engineering Series
this is the *“how to operate OBS”* layer—so the system can *know when it doesn’t know* and repair over time.
Organizations
None yet