ValTEx (Valence-Telemetry Exchange)

Description

A metric combining emotional valence with behavioural telemetry to detect drift in human-AI trust, used to steer conversational AI product strategy.

ValTEx is listed among Vivek Gopinath's selected contributions as a metric to detect human-AI trust drift that influenced conversational AI product strategy. The sources give only this one-line definition; the protocol below is a reconstruction consistent with his drift and DAG-SEM work rather than a published specification. The idea implied by the name is an exchange between stated or expressed valence (how users feel about the AI) and telemetry (what they do with it, such as accepting suggestions without double-checking), so that trust drift is detected when the two diverge or decline together. This parallels the Credible proxy in his drift playbook (adopting recommendations without double-checking) and the trust → double-checking path in the payments SEM.

Attributes

Loss type
Community-proposed
Stage
Post-launch
Data
Mixed
Effort
Weeks

Steps

  1. Define trust-relevant valence measures (in-conversation sentiment, short trust prompts) and trust-relevant behaviours (acceptance without edits, re-asking, verification actions).
  2. Collect both streams per user over time.
  3. Normalise and align the streams per session.
  4. Track their joint trend and divergence to flag trust drift.
  5. Probe flagged users with interviews to name the mechanism.
  6. Feed findings into AI product decisions and re-measure after changes.

Sources

  • Vivek Gopinath, professional research profile (Selected Contributions)
  • Vivek Gopinath (2025), Detecting Behavioral Drift: A Mixed Methods Playbook and E-commerce Case Study

Also known as

  • ValTEx
  • Valence-Telemetry Exchange

Categories

  • Community-Proposed Frameworks