On March 6, Anthropic released its latest research report, introducing a new metric called 'practical exposure' that incorporates real-world application data from the Claude large language model. The report notes that while AI's practical applications are still far from reaching their theoretical limits, hiring for young professionals in high-exposure occupations such as programming and customer service has significantly slowed. Traditional forecasts have largely been based on theoretical capabilities, whereas the new metric incorporates automation trajectories from actual work scenarios. Data shows that 74.5% of computer programming tasks, 70.1% of customer service representative tasks, and 67.1% of data entry tasks are now automated. Although highly educated, high-income white-collar workers are more heavily impacted by AI, no mass unemployment has occurred yet. The study found that positions held by senior professionals remain relatively stable, with some even seeing increased productivity due to AI assistance. However, companies are adjusting their recruitment strategies by reducing hiring for entry-level positions, affecting career entry points for young job seekers. The hiring rate for individuals aged 22 to 25 in high-exposure occupations has declined, with a roughly 14% drop in job search success rates.