The Age of A.I. and Our Human Future

by Henry Kissinger

Most beguiling, though, is what the AI was able to identify. Chemists have devised concepts such as atomic weights and chemical bonds to capture the characteristics of molecules. But the AI identified relationships that had escaped human detection — or possibly even defied human description.

The AI did not just process data more quickly than humanly possible; it also detected aspects of reality humans have not detected, or perhaps cannot detect.

AI will usher in a world in which decisions are made in three primary ways: by humans (which is familiar), by machines (which is becoming familiar), and by collaboration between humans and machines (which is not only unfamiliar but also unprecedented).

The language-generating AI GPT‑3 has demonstrated the ability to create synthetic personalities, use them to produce language that is characteristic of hate speech, and enter into conversations with human users in order to instill prejudice and even urge them toward violence.7 If such an AI were to be deployed to spread hate and division at scale, humans alone may not be capable of combating the outcome. Unless such AI is arrested early in its deployment, manually identifying and disabling all its content through individual investigations and decisions would prove deeply challenging for even the most sophisticated governments and network platform operators. For such a vast and arduous task, they would have to turn — as they already do — to content-moderation AI algorithms.

The push and pull of individuals, companies, regulators, and national governments seeking to shape and channel AI‑enabled network platforms will grow increasingly complex, conducted alternately as a strategic contest, a trade negotiation, and an ethical debate. Questions that appear urgent may be out of date by the time the relevant official participants have gathered to discuss them. By that time, the AI‑enabled network platform may have learned or exhibited new behavior that renders the original terms of the discussion obsolete or insufficient.

Once employed in a military conflict, the technology’s speed all but ensures that it will impose results at a pace faster than diplomacy can unfold. A discussion of cyber and AI weapons among major powers must be undertaken, if only to develop a common vocabulary of strategic concepts and some sense of one another’s redlines. The will to achieve mutual restraint on the most destructive capabilities must not wait for tragedy to arise.

AI will expand what we know of reality. It will alter how we communicate, network, and share information. It will transform the doctrines and strategies we develop and deploy.

The boundary between humans and AI is strikingly porous. If children acquire digital assistants at an early age, they will become habituated to them.

The irony is that even as digitization is making an increasing amount of information available, it is diminishing the space required for deep, concentrated thought.

If an AI is instructed to scan a century’s worth of music or television and produce “a hit,” does it create or merely assemble? How will writers, actors, artists, and other creators, whose labors have traditionally been treated as a unique human engagement with reality and lived experience, see themselves and be seen by others?

Much like the statesmen of 1914 failed to recognize that the old logic of military mobilization, combined with new technology, would pull Europe into war, deploying AI without careful consideration may have grave consequences.

Or does the fact that the AI acted sever it from its creator, at least in terms of culpability? If AI is enlisted to monitor signs of criminal wrongdoing, or to assist in judgments of innocence and guilt, must the AI be able to “explain” how it reached its conclusions in order for human officials to adopt them?