WILL WE RECOGNIZE IT WHEN IT HAPPENS?

BEATRICE GOLOMB

Professor of medicine, UC San Diego

广告：个人专属 VPN，独立 IP，无限流量，多机房切换，还可以屏蔽广告和恶意软件，每月最低仅 5 美元

Many potential paths lead to a technological superintelligence, onto which a supremacy imperative can be affixed—a superintelligence that might enslave or annihilate humankind.

Technology has long outstripped humanity across legion competencies—even many for which evolution designed us. For instance, discerning the sex of human faces is a task we’re designed by evolution to do. Yet even there—and already a quarter century ago—computers bested us. This is one among illimitable illustrations that for myriad tasks, computers have long eclipsed humans. Since those primordial days, countless innovations and applications (GPS, drones, deep networks) provide pieces of a puzzle that, when interconnected, offer a profusion of paths toward future extermination or domination of humans by machines. But just as the target for computer “intelligence” shifts as we acclimate to the latest ability, so too the march toward technological supremacy may go unnoticed, as each incremental encroachment is taken for granted.

Must we even await the future? The answer depends how one defines the question.

1.  How many steps removed must the human input be, to deem the technology culpable?
2.  How clear must the chasm be between machine and human?
3.  Must malice aforethought drive humanity’s destruction or subjugation?
4.  Must everyone be killed or enslaved?

Even insisting upon actions far removed from human input, proscribing human/computer fusion (or collusion!), prescribing premeditation and mandating that all humankind be massacred: The potential remains clear. But suppose we relax these constraints?

1.  If human input need be at no remove: Fretting over whose finger was on the proverbial button (as enshrined in Tom Lehrer lyrics) predates the rise of modern digital technology.
2.  Capacity-enhancing wearables/externals (from old-fashioned glasses and ear trumpets to hearing aids, iWatches, and Oscar Pistorius legs) and implantables (cochleas, pacemakers, radio-controlled spinal devices for paralyzed persons) blur the partition between human and machine. The keen and reluctant alike partake, invested with child-finder microchips or adorned with GPS ankle bracelets. Primitive exemplars have long flaunted their destructive potential—explosives belts as wearables, biological-warfare agents (like the smallpox deployed to vanquish Native Americans) as implantables.
3.  May humanity’s downfall be epiphenomenal? Or must technology “premeditate” human death, decline, or subjugation?

Regarding subjugation: Many now devote their existence to serv(ic)ing technology and nurturing its “evolution.” They mine minerals, design devices, craft programs and apps, or abet devices’ diaspora—channeling custody to caregivers who can serve and service them—or paying for same. (Humans service technology, enabling technology to better conduct “its” business, even as technology services humans so we can better conduct our own.) A dispassionate onlooker could justly ask which is master and who is slave.

Regarding Death, Decline, Disability: TICs and TIMs (toxic industrial chemicals, toxic industrial materials)—from production, use, distribution, and disposition of technology—and electromagnetic exposures (from technology itself, or communication signals therefrom) contribute to the explosion of oxidative stress (OS—cell injury of a type against which antioxidants protect) and associated human afflictions (cancer, neurodegenerative disease, obesity, metabolic syndrome, autoimmune disease, chronic multisymptom illness, autism spectrum disorder).

The lattermost conditions seem selectively to attack the best and brightest—the would-be superintelligent?—of our own kind, as others also observe. I suggest that since OS injures mitochondria, the energy powerhouses of cells, and since those whose biology disposes them to greater brain connectivity and activity also demand more cell energy, such potentially superpowered persons have a heightened risk of cell damage and death. Shortfalls of energy supply due to OS are magnified in settings of high energy demand. (Even “typical” human brains, at about 2 percent of body weight, consume about 20 percent of the oxygen and 50 percent of the glucose of the total body.) One consequence: The rise of superintelligent computers may already have come at selective cost to the would-be superintelligent among us.

So, yes, in the obvious sense, technology may become superintelligent and elect to annihilate or enslave us. But it may progress to similar ends through less obvious means—and may be in that process as we speak.