AI accelerates college entrance exam application process; Baidu completes the "expert backup" puzzle

```

In the past, candidates and parents had to search for information such as historical cutoff scores, university rankings, program introductions, and employment prospects on their own when filling out their college application preferences, then compare them against scores and rankings.

With the intervention of AI, new possibilities are emerging for college admission preference submission.

On June 10, Baidu released a comprehensive upgrade plan for this year's college entrance examination services and launched a new AI preference report.

Specifically, the process mainly involves the Wenxin Assistant having multiple rounds of conversation with the candidates to collect information such as their province, score, interests, preferred majors, and family resources, forming a personalized profile. The system then incorporates multi-dimensional databases including past submitted scores, university and program data, as well as employment prospects for programs, to generate a preference submission plan for the candidates.

Moreover, Baidu’s AI preference report does more than simply list universities and programs available to candidates; it also provides the logic of the recommendations, explaining why these schools are suggested, how to differentiate between grades of preferences, which choices carry risks, and the future employment prospects for relevant programs.

At the data foundation level, Baidu has integrated two types of core data for this upgrade: First, user search data reflecting actual decision-making behavior; Second, industry employment big data co-built with multiple authoritative institutions, covering more than 2,000 majors and focusing on graduates and newcomers to the workplace, helping candidates understand the relationship between program choices and future employment.

However, college entrance exam preference submission is a scenario with extremely high requirements for accuracy and trustworthiness. Although AI can enhance the efficiency of information integration, issues such as AI hallucinations, misjudgment of marginal scores, and misinterpretation of program information still require additional mechanisms for calibration.

This time, Baidu’s AI preference submission process introduces an expert endorsement mechanism, which is freely open to candidates.

According to Jiang Ning, Baidu’s head of education business, speaking to Wallstreetcn, Baidu’s AI preference report expert endorsement mainly has three layers of meaning:

First, expert experience is incorporated during model training. Baidu uses experiences from preference submission experts in real consultations as important corpora for training the model;

Second, a rule engine is connected after the report is generated. After the AI generates the preference report, it connects to rule engines from partners such as China Education Online. Preference submission consultants break down their experience into rules, which are continuously added to the rule library. Most reports are fine-tuned and calibrated by the rule engine upon generation, thus improving basic accuracy.

Third, special reports undergo manual review. For reports where candidates are on the margin of cutoff scores, have high-risk preferences, or where the system's judgment is uncertain, Baidu will arrange for human experts to review them. If any issues are found, the platform will notify users immediately.

This mechanism also reflects a new trend with AI entering the preference submission scenario: AI is responsible for efficiency and coverage, while expert experience serves as a safety net and calibration, forming a synergy in high-risk decision scenarios.

Nevertheless, no matter how much the AI report is upgraded, it still only serves as an auxiliary tool. College preference submission is a typical high-risk scenario, involving candidates' interests, personalities, and long-term career expectations. No recommendation should be simply regarded as a final answer.

For candidates and parents, AI-generated preference plans still need to be cross-verified with admission regulations, historical admissions data, official university information, and actual conditions of programs. In key areas such as cutoff score margins, major adjustments, and special admission requirements, especially, cautious judgment is needed.

Risk Warning and DisclaimerThe market carries risks; investment requires caution. This article does not constitute personal investment advice and has not considered the individual investment goals, financial status, or needs of specific users. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular circumstances. If investing accordingly, responsibility is borne personally. ```