serban·BOB·HAI·PSR·ToT

Comparison

HAI vs. Other Approaches

HAI isn't the only way to handle AI transparency. Here's how it compares to the most common alternatives. The goal isn't to claim superiority — it's to help you choose the right approach for your context.

Comparison Matrix
DimensionSimple DisclaimersNo AttributionAI WatermarkingHAI Framework
Granularity
Low
None
Medium
High
Per-Deliverable Tracking
Pricing Model Support
Open Source
Spec Quality Tracking
Retrospective / Learning

Detailed Analysis

Simple Disclaimers
“AI-assisted” or “Created with AI tools” — a blanket statement applied to all work. Common in content creation and consulting.

Binary attribution (used AI or didn’t). No information about who led the work, how much human judgment was involved, or what the AI actually did. Satisfies minimum disclosure requirements but tells stakeholders almost nothing useful.

No Attribution
Most common approach today. No disclosure of AI involvement. Clients assume all work is human-produced.

Works until it doesn’t. As AI capabilities become more visible, clients will start asking. Having no framework means scrambling to explain when the question comes. Regulatory pressure is increasing in EU markets.

AI Watermarking
Technical approaches to marking AI-generated content (C2PA, Content Credentials). Focus on proving provenance of generated artifacts.

Solves a different problem: provenance of generated content. Useful for images, video, and text where “was this AI-generated?” matters. Does not address the collaboration question: who specified, who produced, how much human judgment was involved. Complementary to HAI, not competitive.

HAI Framework
Per-deliverable tracking of who led, how much human attention, spec quality, and what each side contributed. Open methodology.

Tracks at the work item level: who specified, who produced, what the spec quality was, how much human judgment was involved. Feeds directly into sizing and pricing. Retrospectives compound learning. Works with any tools. MIT-licensed framework.

When to Use What

You just want to comply with disclosure requirements

Simple disclaimers are sufficient. Add “AI-assisted” to your deliverables.

You produce AI-generated images or video and need provenance

AI watermarking (C2PA/Content Credentials) is the right tool. HAI is complementary but doesn’t solve provenance.

You want clients to understand what they’re paying for

HAI. Per-deliverable tracking with size-based pricing gives clients real transparency. The framework connects AI involvement to value.

You want to improve your team’s specification skills

HAI. The spec quality dimension and retrospective cycle are specifically designed for this. No other approach tracks specification quality.

You want to track partner contributions in shared-stake work

HAI Extended. The framework integrates with contribution tracking (Slicing Pie model) — every work item records who specified and who produced.

Note

We welcome other frameworks and approaches. More transparency in AI work is better for everyone — clients, practitioners, and the industry. If HAI isn't right for your context, use what works. The important thing is that you're asking the question.