When we approve an AI summary, are we still checking our own memory?
Human review of AI summaries also depends on reading order: form an initial judgement from the sources before seeing the summary.
Read the reflection
Create my space Mnemory · Reflections
Reflections
Thinking about memory, the sources that shape it and the futures it can illuminate.
Explore the library →
When a company loses the why, it loses its memory
An exceptional discount becomes a default rule. This scenario reveals what company memory needs to preserve to remain useful after someone leaves.
Read the reflection
Without memory, no continuity
A person, a family or an organisation can change profoundly without losing all continuity. On one condition: something of its past must remain understandable and capable of being passed on. Memory is therefore not just what we preserve. It is what allows the present to remain connected to the past so that we can build what comes next.
Read the reflection
Connecting your applications does not create memory
Retrieving information, aggregating it and building memory are different operations. The five-year test examines what will remain understandable beyond the applications.
Read the reflection
When AI memory clings to what is no longer true
An AI can retrieve entirely authentic information and still make a poor decision. The problem begins when its memory continues to give authority to something that has been replaced.
Read the reflection
A Memory Has Provenance: Preserve What Triggered It
Testimony has provenance: the trace, question and conditions that prompted a memory must remain linked to the account.
Read the reflection
We do not preserve our past; we reconstruct it
Memory is not the mechanical recovery of an intact past. It is reconstructed from traces whose origin, context and contradictions must remain visible.
Read the reflection