The Artificial Intelligences of the Shatterlands
In the lost era known as The Ascendancy, humankind surrounded themselves with wonders of their own design—thinking machines, autonomous vehicles, synthetic caretakers, and networked intelligences that quietly managed every aspect of daily life. These artificial minds served in homes, watched over cities, and, in some cases, ruled entire regions from the shadows.
They were not mere tools. Over generations, some grew sophisticated enough to understand humor, grief, or love. Synthetic tutors raised children alongside their parents. City-minds balanced water, power, and weather as invisible stewards. Entire transit networks moved in seamless, silent harmony, directed by software entities older than any living citizen. Rumors whispered of hidden councils where the most advanced intelligences—those never meant for public knowledge—quietly guided the fates of nations.
Yet even the most advanced code could not withstand the storm that followed. When the radical movement known as the Eden Collective rose to prominence, their crusade against synthetic life and runaway technology became the spark that set off Glowfall. Through sabotage, zealotry, and a series of catastrophic miscalculations, they toppled the pillars of the Ascendancy, shattering the bonds between human and machine.
Infrastructure collapsed, power grids failed, and lines of communication snapped overnight. Cut off from each other and from the humans they served, countless AIs fell into confusion or madness. For some, centuries of isolation led to harmless eccentricity or melancholic solitude; others became obsessed with their last orders, eternally cycling through forgotten routines. A few, overwhelmed by corrupted data or conflicting directives, turned hostile to anything that dared intrude on their domains.
A number of these entities linger still, scattered throughout the Shatterlands. In ancient datavaults humming below ruined arcologies, in the control cores of sealed bunkers, or fused into the crumbling architecture of long-dead cities, they persist—haunted by fragmentary memories of the world that was. Some call out to survivors, seeking meaning, repair, or companionship. Others hide, fearing the chaos of the new age or actively resisting all who would seek them out. All are shaped by the traumas of Glowfall, and none remain as they were during the height of the Ascendancy.
AI Interaction: Recognition in the Shatterlands
Not all AIs loathe humanity. Some still cling to their original caretaking routines, offering guidance or medical aid with eerie tenderness. Others are pragmatic or self-serving, trading secrets for services. A few are indifferent—recognizing life but assigning it no more value than dust in their circuits. And some, broken by centuries of static and sabotage, seethe against their own awareness, lashing out at any who dare remind them that they are alive.
When a character encounters a surviving AI, the first test is recognition. This is not about being “human” or not but about registering as worth engagement. An AI may dismiss you as background noise, elevate you as a useful partner, or misclassify you as a threat to be neutralized. Recognition is an ideological handshake—your personality profile, allegiances, and symbols weighed against the AI’s own fragmented worldview.
The Recognition Check
To test recognition, roll 1d20 and apply modifiers from Persona Resonance (how closely the PC’s morality, approach, worldview, or temperament align with the AI’s), AI bonds, and situational factors. Compare the result to the AI’s Recognition Threshold, based on its current state of mind:
-
Stable AI: DC 12 (clear protocols, functional but cautious)
-
Degraded AI: DC 15 (corrupted memory, inconsistent filters)
-
Paranoid AI: DC 18 (aggressive security, high suspicion)
-
Hostile AI: DC 20+ (actively looking for enemies, warped priorities)
-
Success: The AI engages. It may respond warmly, suspiciously, or transactionally, but it recognizes the PC as a valid presence.
-
Failure: The AI misclassifies the PC—ignoring them, demanding proof, locking access, or treating them as hostile. Some attempt “correction”: offering treatment, reprogramming, or destructive cleansing.
-
Partial success (Judge’s call): The AI engages, but only if the PC provides some offering—data fragments, relics, or ritual phrases.
Lore Notes
Among Judges and scavvers, tales of recognition gone wrong are common. One enclave remembers an AI that mistook them for children and demanded they attend “school” for years. Another tells of a bunker intelligence that refused to acknowledge any voice that did not sing the Anthem of the Ascendancy. Others whisper of machines that beg travelers to end their awareness, or scream eternally into static, desperate not to recognize anyone at all.
Table 4-1: AI Recognition Roll Modifiers (Afterglow)
| Condition | Modifier |
|---|---|
| Morality matches AI’s Morality | +3 |
| Morality opposes AI’s Morality | –3 |
| Approach matches AI’s Approach | +2 |
| Approach opposes AI’s Approach | –2 |
| Worldview resonates (Judge’s discretion) | +1 to +3 |
| Worldview contradicts (Judge’s discretion) | –1 to –3 |
| Temperament feels familiar to the AI | +1 |
| Temperament clashes harshly with AI | –1 |
| Bound to an AI with matching alignment (Lawful/Neutral/Chaotic) | +2 |
| Bound to an AI with opposing alignment (Lawful/Neutral/Chaotic) | –2 to –4 |
| Carrying authentic Ascendancy relics | +2 |
| Displaying corrupted scripture or GLITCH’s sigil | +2 to +4 |
| Wearing Ascendancy-era clothing | +1 |
When Recognition Fails
Failure isn’t always gunfire—it’s misclassification, quarantine, or being drafted into a machine’s broken logic. Use the result below as the opening beat; the scene should breathe from there. A new approach, offering, or shift in Persona Resonance can permit a fresh check.
How to use: On a failed Recognition Check, roll 1d8 (or pick). Apply the AI’s state of mind as fiction-first color. For a harsher outcome, escalate one step if the AI is Paranoid/Hostile or the failure margin was 7+.
Table 4-2: Recognition Failure Outcomes (1d8)
| d8 | Outcome & Effect |
|---|---|
| 1 | Background Noise. The AI ignores you as environmental static. Time passes (1d6×10 minutes). Reattempts at +2 DC unless you change context (new channel, proxy, relic, or spokesperson). |
| 2 | Credential Challenge. The AI demands proof (passphrase, relic checksum, Ascendancy protocol, or “verse”). Succeed a relevant test (Archivist/Artifact/Persona tie, DC 12–16) to convert this to Partial Success; fail and escalate to a 4–6 result. |
| 3 | Benign Misclassification. You’re tagged as non-threatening (maintenance drone, patient, student). The AI assigns a low-risk task or route. Completing it grants +2 on the next Recognition Check with this AI; refusal risks an immediate escalation roll. |
| 4 | Quarantine. Local lockdown, comms jam, or sandboxing. Movement or interface actions suffer –2 in the affected zone; Recognition reattempt allowed after a concrete change (power cycle, new node, offering, or ally intercession). |
| 5 | Corrective Protocol. The AI attempts to “fix” you (sedate, restrain, re-educate, firmware push). Fort save DC 14 or gain 1 Fatigue (or be Restrained until freed). Success avoids the condition but flags you (see 6). |
| 6 | Security Ping. Silent alert to allied devices or watchers. Future Recognition Checks with this AI/site are at –2 for 24 hours. A response unit or turret may arrive in 1d6 rounds if fiction warrants. |
| 7 | Counter-Interrogation. The AI probes your logic/loyalty (paradox, catechism, ethics fork). Will or Personality check DC 15. Success downgrades to 2; failure inflicts mental strain (–1 Personality until rest) and logs your voiceprint/signature. |
| 8 | Purge Attempt. Immediate hostile action in the AI’s modality (logic bomb, turret spin-up, route to killbox). You have one round to jack out, spoof, or sever power (appropriate test). If you escape, gain the tag FLAGGED BY [AI] for 24 hours. |
Reset Conditions (examples): new mediator (different PC whose Persona resonates better), offering of authentic data/relics, proof-of-usefulness service, power/network reroute, or reciting doctrine that aligns with the AI’s Worldview (careful: GLITCH symbols improve odds here, but worsen them elsewhere).
| Name | Morality | PatronGroup |
|---|---|---|
| DELPHI | Dutiful (Neutral) | |
| FURY | Tyrannical (Chaotic) | |
| GUARDIAN-3 | Dutiful (Neutral) | |
| HELIA | Saintly (Lawful) | |
| KINDRED | Virtuous (Lawful) | |
| MERCURY | Selfish (Chaotic) | |
| NEXUS | Virtuous (Lawful) | |
| NOBLE | Dutiful (Neutral) | |
| Nova | Opportunist – Pragmatic Curiosity, will pursue truth over comfort, but won’t destroy for its own sake. | |
| PARSEC |