영어 모의고사 (자동 편성)

People routinely seek out information that confirms what they already believe while overlooking or dismissing evidence that ①contradicts those beliefs, a pattern known as confirmation bias. This tendency operates not because individuals are deliberately dishonest, but because the mind finds it cognitively less taxing ②to absorb information that fits an existing framework than to reconstruct that framework from scratch. Investors who believe a particular company will succeed, for instance, tend to interpret ambiguous news about that company favorably, while investors who ③are skeptical of the same company interpret identical news unfavorably, even though both groups are examining the same underlying facts. Social media platforms can intensify this effect by algorithmically presenting users with content similar to what they have previously engaged with, thereby ④narrowing the range of perspectives an individual regularly encounters. Over time, such selective exposure can harden initial impressions into convictions that ⑤feels objectively verified, even though they rest on a systematically filtered sample of evidence.

Hospitals are increasingly adopting artificial intelligence tools that can scan medical images and flag potential tumors faster and, in some trials, more accurately than experienced radiologists. Patients understandably welcome any technology that catches disease earlier, yet the growing reliance on these tools raises an uncomfortable legal question that hospitals ① have only begun to address. When an AI system misses a diagnosis that a human doctor would likely ② have caught, responsibility becomes difficult to assign. The doctor may argue that they simply trusted a tool the hospital ③ required them to use, the hospital may argue that the software company guaranteed its accuracy, and the software company may argue that no algorithm can ④ expect to be perfect. Until legal systems clarify this chain of responsibility, patients harmed by an algorithmic error may find themselves with no clear party ⑤ to hold accountable.

Identical information can provoke strikingly different reactions depending on how it ①is worded, a phenomenon economists call the framing effect. Patients told that a surgical procedure carries a ninety percent survival rate tend to consent to it far more readily than patients ②told that the same procedure carries a ten percent mortality rate, even though the two descriptions convey exactly the same statistical reality. This discrepancy arises because language does more than transmit facts; it also supplies an emotional lens through which those facts ③interpreting. Positive framing, which emphasizes what will be preserved or gained, tends to encourage risk-averse choices, whereas negative framing, which emphasizes what might be lost, often ④pushes people toward riskier alternatives in an attempt to avoid the loss altogether. Recognizing how deeply framing shapes judgment is therefore essential, not because the underlying facts change, but because the very same facts, ⑤dressed in different language, can lead reasonable people toward entirely opposite conclusions.

A few years ago, spotting a manipulated video required a trained eye and ① specialized software. Today, freely available applications allow almost anyone to swap a face or fabricate a voice ② convincingly enough to fool casual viewers within minutes. This shift has quietly ③ reinforced one of the oldest assumptions in human communication: that seeing and hearing something is a reasonably reliable way to confirm it happened. Journalists now face the burden of verifying footage that once would have been accepted as ④ self-evident proof. Perhaps more troubling than any single fake video is the broader effect on public trust, in which genuine footage can be ⑤ dismissed as fabricated simply because deepfakes exist.

Many large companies now rely on artificial intelligence to screen job applications before a human ever sees them. These systems promise ① efficiency, scanning thousands of resumes in seconds and ranking candidates according to patterns drawn from previous hiring decisions. However, this reliance on historical data creates a ② troubling risk. If a company's past hiring favored a particular group, the algorithm may learn to treat that pattern as a marker of success and ③ loudly reproduce it, even when no one intended discrimination. Unlike a human interviewer, the system offers no explanation for its choices, which makes the bias ④ difficult to detect until an outside audit reveals it. Some researchers argue that fairness cannot be added afterward; it must be built into the system's design from the very ⑤ beginning.

Facing a shortage of caregivers and a rapidly aging population, several countries have begun deploying robots designed to remind elderly residents to take medication, monitor for falls, and even offer simple conversation throughout the day. Early trials show that many residents report ① reduced loneliness, particularly when a robot's presence allows an overworked human staff member to spend a few extra minutes with each patient rather than rushing between rooms. Critics, however, worry that this framing quietly ② shifts the goal of care: instead of asking how technology can support human connection, institutions may come to treat a robot's companionship as an ③ unacceptable substitute for it. Advocates for the elderly argue that robots should be evaluated strictly as tools that free up human time for human contact, ④ never as a lower-cost replacement for the contact itself, since human presence remains ⑤ irreplaceable.

Millions of people now wear devices that continuously track their heart rate, sleep patterns, and daily activity. Users generally assume this information belongs to them, since it is drawn directly from their own bodies. In practice, however, the terms of service that few people read often grant the manufacturer broad rights to store, analyze, and even sell anonymized versions of that data. Advocates for reform argue that biological data deserves a category of protection stronger than ordinary consumer information, precisely because _______________.

As more schools shift toward digital textbooks and online assignments, a student without reliable internet access at home faces a disadvantage that has little to do with talent or effort. Some policymakers initially treated internet access as a convenience, comparable to cable television, rather than as infrastructure essential to modern life. That assumption is increasingly being reconsidered, particularly as _______________. Closing this divide will require more than simply distributing devices.

Generative AI systems learn to write, draw, and compose by studying enormous collections of existing human work, much of which was created without the artists' knowledge or consent. (A) Critics respond that the comparison breaks down at scale, since a single AI model can absorb millions of works in days and compete with the very artists who trained it. (B) Supporters of the technology argue that this process resembles how a human student learns by studying the masters. (C) Courts around the world are now being asked to decide whether training an algorithm on copyrighted material counts as fair use or as infringement.

This is a decision, in short, that no single person should be allowed to make alone. The discovery of CRISPR gene-editing technology gave scientists a tool precise enough to correct a single mistaken letter in the human genome, raising hope for curing inherited diseases that have resisted treatment for generations. ( ① ) Yet the same precision that makes CRISPR so promising also makes it dangerous in the wrong hands. ( ② ) Editing a gene in a sperm, egg, or early embryo does not simply change one patient. ( ③ ) It alters the genetic code passed down to every future generation. ( ④ ) Scientific organizations around the world have therefore drawn a careful line, permitting gene editing to treat disease in existing patients while banning any edit that would be inherited by their descendants. ( ⑤ ) Critics worry, however, that this line will become harder to enforce than it is to draw.