After reviewing Microsoft's productivity research and AIIC's ethical stance- one thing is clear: interpreters need practical tools to navigate the AI landscape- not just criticism or cheerleading. Here's a framework built for daily use.

Why We Need a Framework (Not Just Opinions)

The research gives us two complementary lenses:

Microsoft's data shows us what happens when AI enters workflows (productivity gains- workslop- social penalties)
AIIC's statement tells us what we should demand (transparency- consent- high-risk safeguards)

But neither gives interpreters a field-ready checklist for the moment they're asked to use- or consider using- an AI tool in preparation- during an assignment- or as part of platform workflows.

The 5-Question Framework

1. What is the tool actually doing? (CAI vs. AST distinction)

Is it Computer-Assisted Interpretation (CAI) — supporting you as the interpreter? (e.g.- lookup- note management- feed control)

Or is it Automated Speech Translation (AST) — attempting to replace you entirely? (e.g.- direct-to-audience captions or voice)

Why it matters: AIIC is clear these require different standards. Confusing them leads to inappropriate use in high-stakes settings.

2. What are the known failure modes for my language pair and content type?

Does the provider disclose: hallucination rates? bias patterns? performance drops in low-resource languages or specialized domains (legal/medical)?

Microsoft's workslop research shows 40% of users received low-quality AI content from colleagues in the past month- much of it undetected.

Action: Request specific error rates per language/domain. If they can't provide it- assume high risk.

3. How does this tool affect trust and perception?

Consider: Will using this tool (even for prep) trigger the social penalty Microsoft documented — being seen as lazier or less competent?

More importantly: How will clients or Deaf consumers perceive its use?

Action: Test disclosure. Does explaining your AI use build trust- or erode it? Adjust accordingly.

AIIC's non-negotiable: voluntary use with the right to choose a human interpreter.

For interpreter-facing tools (CAI): Do you have meaningful opt-out without penalty?
For consumer-facing tools (AST): Is informed consent obtained before use?

Action: Draft a 30-second disclosure script for consumers. Practice it.

5. Who benefits if I adopt this tool- and who bears the risk?

Microsoft found productivity gains aren't evenly distributed — often individuals gain while teams see no improvement.

AIIC highlights that AST tools frequently perform worst for marginalized language communities- shifting risk to those least able to absorb it.

Action: Map the stakeholders. If the answer is "the agency saves money and I take on quality risk-" reconsider.

How to Use This Framework

This isn't meant to be a theoretical exercise. Keep it literal — print it- save it as a phone note- or add it to your interpreter's go-bag.

Before adopting a new tool: Run through all 5 questions. Document your answers.
When evaluating platform changes: Use questions 1- 2- and 5 to assess agency-proposed tools.
In team settings: Share your framework. Microsoft's research shows bottom-up sharing drives real adoption gains.
With clients/consumers: Use questions 3 and 4 to guide disclosure conversations.

Why This Matters for Our Work

Interpreters have always been the quality gatekeepers in communication chains. We assess source material for accuracy- tone- and intent on the fly — skills no AI can replicate.

This framework doesn't ask us to become AI engineers. It asks us to apply our existing professional judgment- systematically- to a new category of tools that increasingly shape our work environments.

The goal isn't to reject AI. It's to ensure that when we use these tools- we do so with:

Eyes open to limitations and risks
Clarity about who benefits and who bears the cost
Respect for the consent and autonomy of the people we serve

This framework synthesizes insights from: