영어 모의고사 (자동 편성)

After a viewer finishes an episode, a streaming service rarely ① leave the choice of what to watch next entirely up to chance. Instead, a recommendation algorithm compares that viewer's habits with patterns ② drawn from millions of other users who watched similar content, then ranks thousands of titles in order of predicted interest. This approach keeps viewers engaged far longer than a simple list of new releases ever could, which is precisely why streaming companies invest so heavily in ③ refining it. The system is not without drawbacks, however. Because the algorithm favors titles similar to what a viewer already enjoys, it can quietly narrow someone's exposure to unfamiliar genres, directors, or cultures, ④ creating a comfortable but repetitive bubble. Some platforms have begun inserting occasional wildcard suggestions, deliberately unrelated to past viewing habits, in an attempt ⑤ to counteract this narrowing effect.

Turning on automatic captions used to mean settling for garbled text full of obvious errors, but recent speech-recognition models have made same-language captions ① accurate enough that many viewers now leave them on by default, even when watching content in their native language. The technology listens to audio, breaks it into individual sounds, and matches those sounds against a massive library of words and phrases ② to predict what was actually said. For viewers who are deaf or hard of hearing, this improvement has quietly ③ transformed everyday access to video content that once required a human transcriber. The system still stumbles over overlapping speakers, strong accents, and specialized vocabulary, ④ occasionally producing captions that are unintentionally funny. Even with these flaws, most users agree that imperfect captions ⑤ is far more useful than no captions at all.

Open a music streaming app on any given morning, and a personalized playlist is usually already waiting, assembled overnight from songs the algorithm predicts will match a listener's current mood. This prediction relies on far more than genre alone; the system studies skip rates, replay counts, and even the time of day a song is typically played to build a remarkably ① detailed profile of individual taste. Listeners often ② praise the feature for introducing them to artists they never would have searched for on their own, effectively acting as a tireless music scout working around the clock. Musicians, meanwhile, have mixed feelings about this shift in how audiences discover new music. A song that the algorithm decides not to promote may ③ thrive among listeners no matter how well it was written, meaning that an artist's success can depend as much on pleasing an algorithm as on pleasing an ④ actual audience, a dependence some musicians find deeply ⑤ unsettling.

Every day, email providers quietly block billions of unwanted messages before they ever reach a user's inbox. The filter examines each incoming message for suspicious patterns, including certain keywords, unusual sender addresses, and formatting tricks. What makes the system effective is that it keeps learning after launch: whenever a user marks a message as spam, that action becomes new training data, sharpening the filter's future decisions. This creates a constant back-and-forth, since spammers study which messages get blocked and then adjust their wording to slip through again. As a result, _______________.

Browsing an online store today often means chatting with an AI assistant that asks a few quick questions before suggesting a short list of products tailored to the answers. Retailers report that shoppers who use these chatbots tend to complete purchases more often and return fewer items, since the guided conversation helps buyers choose products that actually fit their needs. Some shoppers, however, remain wary of how personal the questions can feel, wondering _______________. Consumer advocates argue that stores should clearly explain what happens to this data, rather than burying the explanation in a lengthy privacy policy that almost no one reads.

Opening most camera apps today reveals a row of filters that can smooth skin, brighten eyes, and reshape a jawline before a photo is even taken, all powered by AI models trained to recognize facial features in real time. (A) Psychologists have grown concerned about a specific consequence of this shift: people increasingly compare their unfiltered reflection in a mirror to a filtered version of themselves seen constantly on a screen. (B) What began as a playful novelty has become a default setting for many users, subtly adjusting nearly every selfie posted online. (C) Some social media platforms have responded by labeling heavily edited photos or banning certain filters that alter body shape.

Rather than replacing human creativity outright, many artists, writers, and filmmakers have begun treating AI tools as an unusual kind of collaborator, one that generates dozens of rough options in the time it would take a person to sketch just one. (A) A film composer might similarly use an AI system to generate a rough melody, then rework it entirely by hand until it fits the emotional tone of a specific scene. (B) A novelist might ask an AI model for ten possible opening lines and then discard nine, keeping only the phrase that sparks an idea worth developing further. (C) What makes this collaboration work, according to many creators, is that the human remains firmly in charge of judgment and taste, using the AI's output as raw material rather than a finished product.

In other words, the two stages appear to serve complementary purposes, one restoring the body and the other restoring the mind. Sleep is not a single, uniform state but a cycle that alternates between two very different modes: REM (rapid eye movement) sleep and non-REM sleep. ( ① ) During non-REM sleep, especially its deepest stage, the brain waves slow down dramatically, the body temperature drops, and tissue repair and growth hormone release reach their peak. ( ② ) This is the period most closely associated with physical restoration. ( ③ ) REM sleep, by contrast, brings rapid eye movements, near-total muscle paralysis, and brain activity that closely resembles wakefulness. ( ④ ) Most vivid dreaming occurs during this stage, and researchers believe it plays a central role in consolidating memories and processing emotional experiences. ( ⑤ ) A healthy night typically cycles through both stages four or five times, with non-REM dominating the early hours and REM periods lengthening toward morning. Disrupting either stage, through frequent awakenings or alcohol use, can leave a person feeling unrested even after a full night in bed.

The tool draws on enormous amounts of data, including flight prices, seasonal weather patterns, and reviews left by previous travelers, to suggest a route that avoids obvious scheduling conflicts. Planning a multi-city trip once meant hours spent comparing flight schedules, reading hotel reviews, and mapping out which sights were close enough to visit in a single day. ( ① ) AI travel planners now compress much of that work into a short conversation, asking about a budget, a travel style, and a handful of preferences before generating a full day-by-day itinerary within minutes. ( ② ) Many travelers appreciate having a solid starting point rather than a blank page. ( ③ ) Still, seasoned travelers warn that these itineraries can feel oddly generic. ( ④ ) They often steer users toward the same popular landmarks that every other AI-generated plan recommends, while overlooking smaller, lesser-known places that a knowledgeable local guide might have suggested instead. ( ⑤ )

Some welcome the tool as a quick way to sketch ideas before refining them by hand. Feed an AI music program a mood, a tempo, and a rough genre, and within seconds it can generate a complete instrumental track, ready to be dropped under a video or used as background music for a small business. ( ① ) The software has studied thousands of existing songs, learning which chord progressions and rhythms tend to sound pleasant together, and it recombines these patterns into something new rather than copying any single source directly. ( ② ) For independent creators who cannot afford a composer or expensive licensing fees, this has been a genuine breakthrough. ( ③ ) Professional musicians, however, remain divided. ( ④ ) Others worry that a flood of AI-generated background music could quietly push human composers out of smaller, lower-budget projects, the very jobs many musicians once relied on to build a career. ( ⑤ )