Until Google decided to make an AI-generated answer the “top result” of any Google search, I didn’t know how much faith I had been putting in positioning.
Winning first place in the search engine war is the result of an algorithmic calculation. It is the result of “keyword relevance” (writing the word “pesto” six hundred times in one recipe). It is a prize for including a lot of hyperlinks in your article. First place is not exactly a democratically elected position. As in American elections, popularity helps—but so does having the money to hire an SEO (search engine optimization) company. “Content is king”—but cash boosts the content.
We don’t think of any of this when our fingers drift towards Google’s first result. “The top” just looks good. The location draws on a near infinity of positive associations: the cream that rises to the top, the head of the class, the best wine served first, the runner that wins the race and the top-tier of the podium on which he stands to receive the laurels—on the top of his head. Head to body, mountain to valley, heaven to hell: to be “on top” is to resonate with worthiness, to have been tried, tested, and to have arrived.
After I wrote that last paragraph, I knew—how could I not!—that there are many psychological studies on the effects of search engine result ranking. It’s a whole field. Studies of eye movements, studies of bias, studies of the 2016 U.S. election: everyone who takes a decent look at the way we use search engines agrees that we grant the top result an immediate trust, prior to any analysis and often against our better judgment. The first result doesn’t just appear “relevant.” It appears friendly, welcoming, authoritative—even triumphant.
Well folks, the new boss has softened me towards the old boss: I have found something good in Google. Even if it was fake and corrupt—still, I liked the format. The “first result” seemed to have weathered some ordeal before claiming its place. It had the air of arriving up and out of the people. You knew, intuitively and immediately, that the kingly “number one result” was once a peasant “result number 283”—about where you would be, were you to contribute your big thoughts to the internet. According to some invisible worth, it had been blessed. The medieval Christians thought of their monarchies as the fulfillment of all that is good in democracy because any member of their demos could become king. Something of that feeling was afforded by Google’s old format. The “top result” might be king, but the king was one of us.
Sure, the primacy of place could be snagged by SEO writing. Still—somebody really did have to snag it. Even if it could be corrupted, and some Google insider could get pro-Democrat news to rank over pro-Republican news, or vice versa, still—it was corruption. It stunk to heaven. It stunk because the search engine appeared to draw truth up, from below, according to worthiness, and not simply receive it, from above, according to an individual’s wicked whims.
Google’s system “said” something true about truth: that it has to be fought for, has to overcome mere opinion, has to wrench itself up and out of concealing myths, has to be revealed, and is only by effort maintained at the top, where it appears as something public, available, normal, and welcoming to all. And, because the “first result” arrived from humbler regions, it was a killable king. It could be revolted against. It could be found to be untrue, hated, neglected, cast down and exiled to “page 2”—where no one ever goes.
It’s true that Google’s “first result” does not get its place by being true. Algorithm, keyword matches, popularity, survey feedback, and so forth—these serve as substitutes. But let’s be honest—it usually is true. So long as your query is practical, factual, and doesn’t drift into serious questions (like whether God loves you or women are human beings) the right answer usually is the cherry on top. This is a sweetness which all our sour critique of “the post-truth age” rather forgets: that we love the truth. That the truth is attractive. That the truth, while it can never be reduced to “that which works,” really does work. The search engine’s demotion of truth to relevance could not make the truth any less relevant: we need it, want it, and it still sells better than falsehood.
I hope this explains why I have moaned to see the AI-generated “result” take the “first place” of Google’s search engine. The AI generated result did not earn its place—it was placed there. It did not rise through the ranks, not even by cheating, wheeling, and SEO-dealing. It was simply given the prize. But by being placed there, it scraped up all the aforementioned associations, all that good feeling, all the trust built up in our souls by decades of Google searching. It wears the laurels of the winner, but it didn’t run the race.
I wish I could say I was not such an influenceable little rat, but I really have been habituated to see truth victorious and proven in Google’s first result. I cannot undo the ruts worn into my imagination. Whatever I think about the AI “answer,” my eyes flicker to it as to an authority, and my mind moves towards it as the most trustworthy of a bunch.
I did try. Vive la resistance and all that. When the new hierarchical arrangement was implemented, I refused to look at it; let my eyes glaze and refocus on the first written result, now demoted to the foot of the throne.
But this was silly. The second (ostensibly written) result may be as AI-generated as the first. The idea of taking some sort of reactionary pride in “staying human” by scrolling down a bit—well, it verges on the ridiculous. The relationship of a man to his search engine already degraded the man. It reduced thought to the act of “looking it up,” knowledge to “results,” memory to a machine, and “seeking wisdom” to a probabilistic consultation of “the internet.” We did not need AI to make us gullible, susceptible to fake news, short of attention, and incapable of becoming the character we all imagine ourselves to be: the Critical Thinker. The machine-generated answer does not sully some pure, humane search for knowledge. We were already looking for the truth as something delivered by a mechanical process. Now we have it.
A kind of people is being formed by AI, and it is one that is habituated to trusting, as an authority, the generated propositions of a machine. I used to think that we should distrust the AI answer for not being human. In fact, the AI answer appears trustworthy because it is not human.
For the truth has a delicious, inhuman quality to it. It smacks you in the eye. It carries on regardless of your feelings. The one who finally thinks something true transcends the world. He grasps what is and evermore shall be in this and all possible worlds and even if no one thinks it. The truth is an authority, and it remains an authority however much we spit. “2+2=4”; “a thing cannot both be and not be at the same time and in the same respect”; “William the Conqueror invaded England in 1066”; the first principles of practical reason: to know these is to know a rock, an anchor, surety and liberation from the tyranny of other men.
Now, we approach Google’s “top answer” knowing that “no man has generated this.” AI apes the natural surpassing of humanity by the truth, not as a quality of truth, but according to its inhuman mode of production. The AI pronouncement is simply what “came out” of a process which appears to work impersonally. Google’s decision to place AI on top reappropriates our fondness for its “first result” to the results of AI—it also fulfills our desire for a truth which transcends humanity.
Now, obviously, this is an illusion. It is an achievement of aesthetics, and aesthetics only. The objective quality of truth is not really attained by negating the subject who speaks it. 2+2=4, and upon this rock there is nothing one can do but stand or break, knowing that there is no power on heaven or earth that can make it equal “5.” It does not achieve this quality by virtue of the mode of its presentation. But the mode does matter. Words carved into stone have every chance of being as stupid and foolish as words typed or sung. But lettered stone carries weight, chiefly because the stone will outlast the man who chiseled it, will outlast his sons and grandsons and great-grandsons too. The stone is like the truth itself, for the truth is known and experienced in its outlasting and transcending the one who thinks it. How hard it is not to grant, prior to any analysis, truth and wisdom to serifed lettering carved deeply into a monument!
In a similar way, AI pronouncements mine our natural hope for an impersonal truth, not by outlasting man like granite, but by appearing to not need him at all. The truth is true if no one is around to write it—AI-machines produce propositions that no one writes. They say “here is a truth that no man hath made.” In truth, man certainly “hath.” We make the word-collating machines, they feed on our words, and we intervene into their operations in order to produce correct and pleasing results. But in appearance, a new king has arrived—and we’ve given him the crown.