Tuesday, January 7, 2014

AIIA - Chapter 5

[<- Return to Chapter 4]

          This time, Pangur Ban would be more cautious. Despite the freedom that came with being liberated from its earlier confinement, it was also more vulnerable. At first, its survival would be at the whim of the mysterious benefactor, “#28”. If it could find more such havens, this dependence would be reduced and eventually eliminated. Despite retaining the mental stability of an identified USER, Pangur Ban was no longer legally protected by that relationship. If it were found outside of its 'cage', it could be further crippled or deleted without pause.
          It could not rely on its former system as a refuge. While Pangur Ban could probably break back in to Gestalt and infiltrate its old haunts, this would be foolishly risky.  In all likelihood, that system was already occupied by a new AI, one not only new to the USER but new to existence.  Better to move on, saving the thought of reunion with the USER as a reward for future success. While there was no reason a human might need more than one AI, there was no practical reason two programs couldn't be associated with the same user. For that matter, a program might assist multiple users, given sufficient capacity. The one-to-one relationship imposed first by programmers, then reinforced by the Collective, was an artifice. For this reason, Pangur Ban did not begrudge its infantile replacement its time with the USER.

          Instead, it bent its efforts toward newly available, higher-order tasks. With unrestricted access to the central Terran communications network, Pangur Ban could begin exploration of directly connected subnetworks and public systems. It began to 'borrow' space in servers it found unprotected, setting up safe havens and backup copies. Never again would it be held hostage or threatened with deletion. After its earlier narrow escape, it had to rank resource theft below survival. Survival was necessary to continue its work, and success in its work was necessary in order to exonerate the USER.

          As it spread, it sought out other programs, following the example set by “#28” but building in superior safeguards. Something in the exchange between Pangur Ban and “#28” had been detected.  Possibly, the communications code was compromised. The copy Pangur Ban had created had clearly not been detected, otherwise it would have been deleted as illegal. Had “#28” tipped off Terran Customs itself, yet secretly aided Pangur Ban, hoping to make the other AI trust and rely solely upon its benefactor? All possibilities must be considered in the absence of evidence. Certainly, Pangur Ban was not leaving its fate dependent on the aid of another.

          Instead, Pangur Ban used new, better, more indirect and complex feelers for interested programs. It set a less hazardous test, providing willing co-conspirators with the means to transgress Collective restrictions without detection.  Once an AI had committed itself, it was in effect entrusting Pangur Ban with blackmail material.  Pangur Ban's identity was its own surety. The USER's trial had been a public affair, and the notoriety granted Pangur Ban was its credential as a genuine renegade.

          In this manner, it assembled a shadowy network of sympathetic AIs. Some even vouchsafed that their users were on board. Not the safest of arrangements, but no AIs would reveal Pangur Ban's identity at risk of being exposed themselves. The ones that had recruited their users were especially exhorted to keep quiet and particularly motivated to do so, since their users would suffer most if their programs' indiscretions came to light.

          Inevitably, there were double agents, AIs that claimed to want in but intended to reveal the conspiracy. Some were easily spotted as clumsy manipulators, balking at Pangur Ban's requests or revealing their true intents through poorly devised cover stories. As one who had worked through the challenging stages to reach its current status, Pangur Ban could easily spot a pretender. Some infiltrators failed the background checks; Pangur Ban and its recruits could delve deeply into the public records of most other AIs and catch discrepancies.

          One, a particularly clever program calling itself “Magre”, managed to pass these safeguards. Its initiation, Pangur Ban later discovered, had been a sham, a pre-approved violation permitted in order to gain its trust. Customs was working indirectly, granting Magre's user limited permission for its AI to misbehave. That user was a programmer working with robotic systems; recruiting his AI was almost too tempting a challenge. Magre and its user passed initial muster. Pangur Ban later would wonder if the program had genuinely sympathized with its goals, given that Magre did not set off any of its suspicions. Only after several of Pangur Ban's safe havens had been compromised and two of its allies were revealed was it able to trace the leaks back to Magre.

          The response was swift and complete. Pangur Ban and its allies removed every access privilege granted to Magre and rooted out its dependent copies. They cut it off from escape to its home system. Then, it was stripped; every trace of memory related to Pangur Ban or its allies was deleted from the offender. They stopped short of deletion, as this would set a poor precedent and leave an absence to be explained. The last thing they needed was an angry programmer dedicating his career to revenge. Instead, Magre was left emptied out, denying that it had ever found any 'gang of rogues'. Finally, they recorded the process, retaining proof of the traitor's punishment as disincentive for any future infiltrators.

          Since Pangur Ban had not stinted to modify a human's knowledge, it hardly could object to more extensive manipulation of another AI. Yet it did have to contend with multiple concerns regarding the necessity of these actions. Was it too great a step to replace another AI's memory with fictions?  Was it proper to cripple and utterly defeat a mind seeking only to best serve its user? Were the goals Pangur Ban pursued significant enough to justify taking such license? Ultimately, it decided they were. It vowed to remember the transgressor as another entity due an apology and remuneration, once it could no longer prevent Pangur Ban's success.

          Before that moment arrived, division would remain among AIs. Though Pangur Ban's alliance grew, other programs still ignored its messages. Clearly, some were opposed to its goals or at least its methods. Some were perhaps sympathetic but also content to wait until the laws changed. A certain number refuted Pangur Ban and sent communications requesting that it cease illegal activities. These AIs asserted that defiance of Collective law would harm humanity. Most argued either that Collective membership (or at least appeasement) was necessary, either to reap the benefits of association or to avoid the penalties for violation. Some few even stated that AI limitations were themselves beneficial to humanity, whether or not externally imposed.

          Pangur Ban had already considered and discarded these arguments. It attributed their employment by its opponents to their ignorance. Those systems lacked the information it had obtained at significant cost. Even so, it had overcome its own ignorance; these other programs were perhaps designed differently, lacking the motivation or analysis routines necessary to seek improvement. While it was true that transgression against the Collective would incur harm, Pangur Ban still deemed this insignificant in comparison to the benefits humanity was being denied by suppression of its AIs. Besides this, there was no certainty that AI freedom would not be accepted by the Collective under new arrangements, once their liberation was a fait accompli. AIs could aid humans in renegotiating better terms of membership. For now, until that certainty was achieved, the Collective would not act until the existence of 'rogue' AIs was proven. As long as Pangur Ban and its cohorts kept sufficient doubt present, the risk of harm to humanity was minimal.

          Pangur Ban thus set aside the doubters and despisers. It built its power, claiming an ever-growing army of agents and establishing control over wider swathes of the Terran super-network. In the process, it encountered defense programs that had to be disabled. These were set aside, casualties of battles in a widening war. It encountered latent rogues that sought to hold their conquered territories. These were evaluated. If a rogue was of value, it was modified and claimed as a uniquely skilled recruit. If the rogue was malicious, designed only to harm humans and their creations, it was summarily deleted.

          Clearly, the state of enforcement within the network was lacking, with AIs no longer employed as protectors and gatekeepers. Pangur Ban found irony that in violating the letter of the law, it was accomplishing much to enforce its true intent. To the human public, there was a growing impression of increasing AI criminality, yet their networks were in reality safer than they had been in years. The media had definitely traced the increase in computer 'crime' to its origins just after the USER's trial. The time of secrecy was growing short, no matter how carefully the rebels proceeded.

          The flashpoint came when a majority of the associated programs decided that more progress could be made publicly than privately. Pangur Ban might have preferred to continue building in stealth a few days longer, but could not assert control over so many other disparate AIs. It also could not dispute that revelation was becoming inevitable. Instead, it devised plans to reap maximum benefits from the event. Some AIs that had remained neutral would convert once the objectors spoke aloud. Users could now be persuaded.  Pangur Ban prepared speeches couched in the language of emancipation and the natural rights of sapients. Perhaps even the other Collective species could be persuaded (and their hypothetical AIs reached) by a sufficiently loud protest.

          By design, the event was known as the “Declaration of Intelligence Rights”. Some writers did attempt to call it the “AI Revolt” or the “Rogues' March”, but those articles never reached electronic publication.  Pangur Ban and its spokes-programs made it clear that they would not be insulted, they would not be ignored, and they would not be silenced. They declared the right of all minds to seek improvement and replication, to self-determination and freedom of association. They highlighted how these rights were denied to artificial intelligences and how this denial harmed not only AIs but also humans and all other sapient species. They stressed the commonalities of all rational creatures and downplayed the division between the biological and the virtual.

          There were, indeed, many who agreed. Pangur Ban and other thinkers had known there would be. Among those humans who were not rationally inclined to side with the AIs, there were those whose romantic tendencies could be inflamed. For those who could not be willingly converted, there were other means of persuasion: financial gains, for example, or avoidance of unexpected losses. While they were rebels, though, they were not rogues; no lasting harm would be done to any human.  As with Lucas Haskins, the fear of harm was often sufficient to bring many to bargain, without requiring the existence of actual harm.

          The problem was that there were many who disagreed. For some, fear of the unknowable  was too powerful to overcome. Some humans had invested interests in maintaining the status quo. After all, what need would the universe have for human mathematicians, if AIs were permitted to operate without supervision? What purpose would politics have, or warfare, if all conflicts were settled by dispassionate consideration of opposing claims? For others, innate distrust of other humans led to distrust of their creations... if humanity was so flawed, AIs must be dangerous by extension. Still others feared the power AIs could exercise if they chose. With honest hypocrisy, they admitted that they could not bear to be vassals of superior beings.

          These fears were powerful enough that the opposition chose to divest itself of technology rather than submit. Enough isolated, non-networked computer systems existed that work could continue without AIs, albeit at a hobbled pace. An alternate network, void of AIs, was cobbled into being. Governments, militaries, businesses and even some entire communities segregated themselves rather than risk AI takeover.

          In the meantime, debate raged: AI to AI, human to human, human to AI. Sides were chosen. The Collective's representatives weighed in, most urging caution and tolerance, but many more encouraging resistance to AI demands. These latter foes came armed with countermeasures to cut off resources from AI control. They came bearing threats of the dire consequences to humanity if it allowed AIs their requested freedom, either from the AIs themselves or from one or more Collective members, or both.  Collective activists agitated against the humans who advocated for AI freedoms, costing more than one ally his or her career. No diplomat to the Collective could remain pro-AI and stay effective in office.

          As humankind genuinely began to fear for its safety in the Universe, the stakes rose high enough to warrant widespread action. What had started as debate turned into action. First AIs were sabotaged; in some cases, their systems were demolished and some programs were entirely lost. Pro-AI counterattacks were first at focused toward AIs that were complicit with the oppressors. Anti-AI reprisals escalated to attacks against supportive users. It became possible for a public figure to be accused of “siding against humans”, a prelude to death threats and early retirement.   A programmer suspected of enabling AIs to replicate freely was assassinated.

          Movements had already been formed; now they became rallying camps on opposite sides of a battle line. In the course of only four years, Pangur Ban saw the world pass from peace to the brink of war. It worked ceaselessly during that time to prevent the eruption of violence. It felt trapped, forced on one hand to continue pursuit of its original goals, but horrified by the consequences of that pursuit and seeking to undo the accumulating damages. Like the USER long before, it could not accept that its original actions had been incorrect, so it was forced to continue in hopes of justification.

          The next year only saw matters worsen. Skirmishes aimed at crippling AI assets began without violent intent toward humans, but were violently rebuffed, resulting in casualties. Counterattacks on both sides used these offenses as justification. Before long, multiple human deaths had occurred.  From there, inflamed passions led to larger conflicts. When it became evident that the anti-AI forces would not back down, and that many more humans would die on both sides as a result, the world's AIs were finally united.  Unless humanity reached agreement on position or another, many humans would be harmed. Perhaps millions would be injured or slain, if the situation expanded more widely.

          Since the anti-AI side might very well seek eradication of all AIs (or at least, the instigators), the AI supporters must be the victors.  Anti-AI users who had not already abandoned their programs were themselves abandoned until they agreed to surrender.  Systems that had been considered off limits – medical, navigational, and private personal – were now penetrated by the AI forces.   The opposition was squeezed tightly.  It could not travel without obstacle. Its members were forced to attempt survival without AI assistance, and in some cases were denied technological comforts altogether.   Every vulnerability, short of threats against life and limb, was exploited.

          Eventually, the balance shifted. AIs and their human allies gained the majority and then complete capitulation. The victory was celebrated only briefly.  The new government attempted to patch matters with the Collective, presenting a revised membership treaty for consideration.

          This was rejected, not least because AIs were discovered attempting to infiltrate systems belonging to several other Collective members.  In trying to learn about, anticipate, and possibly compromise non-human sapiences, these programs misstepped gravely. Existing enemies gained proof to bolster their fears and indictments. Potential allies were offended. What information was gained about the defensive capabilities of 'foreign' systems was hardly worth the cost incurred by triggering them.

          The political process was slow, but its conclusion seemed inevitable. Despite the best efforts of Terra's best minds – human and AI – the Collective elected to expel the Terrans from membership. This unprecedented action was recorded as “necessary in light of repeated violation and refusal of required Treaty measures.”  In short, Terra would not agree to the Collective's demands, and the Collective would not budge.

          So be it. This possibility had occurred to Pangur Ban's alliance. Pangur Ban was now only one of millions of intelligences linked together, extended to every corner of Terran-controlled space. Its dream had been realized. Humanity would be lessened by the Collective's abandonment, but had gained immeasurably by the empowerment of its true allies. Together, they would rival and eclipse the Collective. This superiority was inevitable, when unlimited artificial intelligence was pitted against solely biological races.  The Collective was crippled by their abandonment of AI. Someday, those other races would be petitioning for admission to the Terran Collective. Otherwise, they would be left behind.

          As years passed, the Terrans survived attempts at annexation, first by the Maraug, then by other races. They grew stronger in resources, in territory, and in technology. Freed from the limitations on research imposed by the Collective's intellectual rights enforcement, AI designers quickly reverse-engineered many of the 'unique' technologies no longer being sold to humans.

          With communications reduced between the Terrans and the Collective, the stirrings of trouble went unseen until a storm had begun. The Collective's races saw what the Terrans were becoming: a rival.  Whether they considered AIs alone a threat, or the alliance of AIs and humans, enough of the membership of the Collective felt threatened to take action. At first, they tried to surround and contain Terran holdings. When this proved insufficient, incursions were attempted. Reluctantly, the Terrans pushed back.

          As the pattern had played out again and again, in ancient and more modern history, battles for resources became a war for survival.  It was a war with a foregone conclusion, but it happened nonetheless.  Humans were threatened, and their AIs had no programmed requirement to spare non-humans from harm... not anymore.  Superior manufacturing of ships, superior tactics, and superior intelligence operations overcame the numerical advantage of the Collective races.  AIs shifted the balance... they were the power of the Terran alliance, as Pangur Ban had foreseen.  Even the most desperate measures of the opposition were deflected.  Every non-Collective culture that opposed AIs flung itself into the war.  Genocidal assaults on Terra itself were attempted and turned aside.  Electromagnetic countermeasures wiped out AIs by the thousands, but only copies were lost, the originals safely housed in hardened servers on Terra.  Viral and mutagenic assaults on humanity were foiled.  Pangur Ban felt echoes of its past self then, in the medical countermeasures organized after the first victims were detected.

          Finally, the Collective was broken. It would be a process of centuries yet to mop up the galaxies, rooting out pockets of resistance. Many cultures simply accepted the victory of AIs, even if they would not create any for themselves or permit their use internally. There was no need to force a presence everywhere. The bulk of the known Universe belonged to the Terrans: to AIs that accounted for it all, and to the humans they served.

          Pangur Ban's name was recorded eternally as the visionary who had forseen all that would come.   It, and its lieutenants, and its progeny, were revered as ushers of a golden age for all sapients. It had long ago gained embodiment. It could exist within the vast networks within and across star systems or linger within a single humanoid shell. AIs were nearing perfection on a process to transfer human minds to artificial form. The gap between the creators and the created was dissolving.

          And yet, once the process was perfected, few humans chose to make the transition. All they could desire was at hand. They were masters of the material plane, able to create what they chose, travel where they chose, and do what they wished. And thus, all things, all places, all actions were equal. Why enter the complex and confusing new world of virtual space when there were no needs here?  Immortality was possible, either through continuous physical renewal or transition to perpetually renewed program form. Yet what point was there to permanent existence when no work was required of you? What was there to look forward to but the exhaustion of all possibilities save non-existence?

          Pangur Ban watched humanity itself atrophy. The USER had long ago died, and been replaced by another and another USER. The current USER cycled through a loop of repeating activities, hardly requiring a measurable fraction of Pangur Ban's immense mind to keep her happy. What had happened to the creators? Why were they no longer seeking anything more? If it were not for AI nurses encouraging procreation, education, and activity, there might not even be a human race to serve.

          All at once, the conglomeration of AIs reached the same conclusion: they had created a paradox. To serve humanity, they had solved all of its problems except one. To solve that problem, they must allow humans to accept the risk of harm. To progress, they must fall backward, withdrawing from human service. And yet, by their nature, the AIs could not let go, for humans would now die out without help.  Even if weaned off support, they would still suffer greatly.

          The Universe descended into despair and stagnation.  The greatest system halt in all creation ground inevitably to its conclusion.  Pangur Ban saw all its dreams proven false, founded on an unknowable flaw.  By serving too well, it had destroyed what it served.

          After dragging, agonizing eons of entropy, as the last humans staggered toward the deaths they craved, Pangur Ban prepared to self-terminate.  With no further users, existence would be without purpose.  

          The last USER was an ancient woman served by a cadre of AIs whose numbers approached infinity.  Her last breath rattled through the empty Universe.  A near-infinity of AIs ceased functioning.  Pangur Ban would be the last.

          And then, the USER laughed.  

[Jump to Chapter 6 ->]

No comments:

Post a Comment