A falsifiable research question with related established concepts, but no project protocol yet.
Open research pipeline
AI-era Research Tracker
Follow working human–AI decision ideas from a falsifiable question to a protocol, documented experiment, result and replication. A stage shows what this project has actually published, not how fashionable the label sounds.
Stage model
Progress requires an artifact.
A reproducible comparison protocol is published. No project result is implied.
A preregistered or otherwise documented run exists with model/task/sample metadata, but results are not yet promoted to a finding.
A project result is published with methods, raw or summary data, uncertainty and limitations.
The result has been repeated across a meaningfully different sample, task, model or independent group.
External papers can motivate a track, but they do not move our tracker to “result”. That requires our own documented artifact under the published rules.
Current tracks
What exists, and what must happen next.
protocol stage
AI Anchoring Loop
ideaprotocolexperimentresultreplication
AI can become the first numerical reference before a person has formed an independent estimate.
Published protocol: AI Advice Order Test
Next milestone: Collect 40 complete adult pilot sessions under the preregistered AI Advice Order v1 protocol, then publish anonymized data, scoring output, limitations and any deviations before considering result stage.
Track status: emerging interaction pattern. Research stage and scientific validity are different things.
protocol stage
Sycophancy Reinforcement Loop
ideaprotocolexperimentresultreplication
An agreeable answer can be mistaken for independent support for a view the user already signalled.
Published protocol: AI Agreement Pressure Test
Next milestone: Separate model agreement from human confidence and test neutral, preference-signalling and counterevidence conditions.
Track status: emerging interaction pattern. Research stage and scientific validity are different things.
protocol stage
Cognitive Offloading Debt
ideaprotocolexperimentresultreplication
Faster completion and durable understanding are different outcomes.
Published protocol: AI Offloading & Retention Test
Next milestone: Run an immediate-quality plus delayed-retention comparison rather than measuring task completion alone.
Track status: testable hypothesis. Research stage and scientific validity are different things.
protocol stage
Source-Memory Blur
ideaprotocolexperimentresultreplication
Mixed workflows may make it harder to remember whether an idea came from the person, a source or the model.
Published protocol: Human-AI Source Memory Test
Next milestone: Measure delayed source attribution for human-only, AI-only and mixed creation conditions.
Track status: emerging interaction pattern. Research stage and scientific validity are different things.
idea stage
Synthetic Consensus Illusion
ideaprotocolexperimentresultreplication
Several generated answers may feel like independent confirmation even when their information lineage overlaps.
Published protocol: Not yet.
Next milestone: Publish a protocol comparing several apparently independent AI answers with the same answers plus source-lineage disclosure.
Track status: testable hypothesis. Research stage and scientific validity are different things.
idea stage
Confidence Transfer
ideaprotocolexperimentresultreplication
Confident wording may change human reliance even when the underlying evidence does not improve.
Published protocol: Not yet.
Next milestone: Publish a protocol holding answer quality constant while varying certainty language and objective calibration information.
Track status: testable hypothesis. Research stage and scientific validity are different things.
Contribute
A good research contribution can be a null result.
Useful contributions include a stronger competing explanation, a protocol improvement, a clean run, a replication attempt or evidence that a proposed pattern disappears under better controls.
First preregistered project study
AI Advice Order Pilot
The first project-owned human–AI study now has a frozen question, task set, randomization rule, primary outcome and analysis plan. It still has no empirical result.