Not all evidence means the same thing.
Same student. Same capability. Different evidence context — and possibly a different, equally reasonable conclusion.
AI-supported practice
- Student action
- Used AI to draft an initial recommendation, then revised after George's challenge.
- Evidence produced
- Identified the weak assumption and proposed a stronger alternative.
- Reasonable inference
- Can respond to feedback and revise reasoning with support available.
Team decision, then individual defense
- Student action
- Helped the team agree a recommendation, then defended it alone when George challenged their reasoning.
- Evidence produced
- Explained which parts of the team's decision they stood behind and where they would have chosen differently.
- Reasonable inference
- Can reason about a shared decision in their own words — kept separate from the team's evidence.
Reduced-support assessment
- Student action
- Responded to a Reasonline Assess follow-up grounded in their own submitted work, with less support available.
- Evidence produced
- Defended the recommendation under a newly introduced constraint.
- Reasonable inference
- Can explain and adapt the reasoning under this more bounded condition.
Evidence Context preserves the condition alongside the evidence — it does not rank the columns above by which is “better.” Each supports a different, reasonable conclusion. When Reasonline recommends something, you can open the exact student work behind it, with these conditions attached.
AI-supported work does not have to be either proof or meaningless.
Much of the current conversation about AI pushes educators toward two extremes: that finished work always tells us everything, or that AI involvement makes it tell us nothing. Reasonline rejects both. The useful question is not whether a kind of evidence “counts.” It is what conclusion the evidence can reasonably support.
Supported learning is meaningful. The support still matters — and belongs with the evidence, not hidden from it.
Different conditions can answer different questions.
Supported learning may help answer
- Can the student respond to feedback?
- Can they apply a framework with guidance?
- Can they use AI productively rather than passively?
Another configured assessment condition may help answer
- Can they explain the reasoning under this condition?
- Can they transfer it to a new situation?
- Can they adapt when a constraint changes?
Neither category automatically represents “better” evidence. Reasonline Assess creates the second condition without claiming to have magically proved “independent capability.”
Evidence Context across Reasonline
| Surface | Role |
|---|---|
| In Learn | Record how George was helping (Tutor, Coach, or Critic), what students had to do themselves, and whether the work was individual or part of a team. |
| In Assess | Freeze the AI-use rules that applied when the evidence was produced, so later edits never rewrite history. |
| In Complete | Connect evidence to the same capabilities while the differences remain visible. |
Context also matters for Learning Intelligence: strong supported evidence may suggest a step with less support; weak or conflicting evidence may suggest more practice before a conclusion.
Explore Learning IntelligenceEvidence Context and AgencyThread do different jobs.
Evidence Context records the conditions of the work: assignment rules, permitted assistance, individual or team setting, and the support Reasonline provided. Reasonline AgencyThread interprets evidence of human participation in work that often happened elsewhere: AI conversations, drafts, version history, and sources, read for how the student framed, directed, evaluated, developed, and took responsibility for the work. Evidence Context doesn’t include AgencyThread. Each AgencyThread report records the conditions it was produced under.
Explore Reasonline AgencyThreadEvidence Context comes from the design of the work.
Reasonline does not secretly infer student behavior to preserve context. The relevant context comes from the explicit design of the experience: what support was available, what the student was asked to do, and what condition was configured. If an institution separately uses its own exam-security tools, those are represented honestly and only to their verified status — Evidence Context is not a claim that Reasonline knows everything that happened outside the system.
AI does not make student work useless. It makes context more important.
Explore Reasonline Complete