使用 AI 進行線上即時技術面試：面試官與面試者的誘因
在觀看過相關影片與網路上的示範後（截至2024年7月2日），我對於使用 AI 進行線上即時技術面試的看法如下：Jump to English version
面試的主要審核方向
假設未來工程師的合作對象還是工程師，面試主要審核方向包括：
- 記憶性知識
- 程式熟練度
- 解題思考方式
- 技術細節溝通適性
- 行為問題
現有方案的局限
目前常見的線上作答網站，對於記憶性知識、程式熟練度已經滿足大部分需求。在衡量解題思考方式方面，線上作答網站也能部分滿足需求，因為從程式碼的撰寫可以看出思考方式。然而，技術細節溝通適性和行為問題無法用線上作答網站自動批改來衡量。
技術細節溝通適性的重要性
關於技術細節溝通適性，有兩個主要衡量點：
- 當面試者想出好解決方案，面試者能使面試官理解。
- 當面試者遇到困難時，面試官是否能夠通過提示實質幫助面試者。
滿足這兩點表明面試官與面試者之間契合。要自動化地回答這兩個問題，需要對面試官的技術想法進行建模。然而，如果面試官的替身模型對技術想法的理解與面試官本人存在很大差距，這將不適合評估適性。
企業形象與效率
面試同時也是企業形象的宣傳。只減少企業方的面試成本可能會形成不重視他人時間的形象。如果面試時雙方都開鏡頭，這通常代表著信任。因此，若要使用 AI 進行線上面試，應同時減輕企業方與求職者的成本，這樣可以吸引那些注重效率的求職者加入公司。
提升面試準確性的潛力
目前的討論主要集中在如何使用 AI 模仿面試官的流程以提高效率。然而，我對網路上 demo 所展示的用法持保留態度，因為那並沒有增加找人的準確度，反而可能會破壞企業形象。
面試的最大問題不是花費太多時間，而是需要準確地知道誰適合這份工作。常見的問題是，面試者可能因為緊張或因初次見面而無法充分表達自己的想法，從而錯失機會。如果 AI 能夠輔助引導面試者提供更多面試官所需的信息，將有助於面試官更好地理解面試者。
另一方面，面試官在審核面試者時，有時可能會過於武斷，在未完全理解面試者的想法前就做出不錄用的決定。如果 AI 能夠提示面試官進一步追問或探討值得探究的議題，可能會避免遺珠之憾。
最後，我希望 AI 能夠提升面試的準確性，不是僅僅追求效率，而是通過輔助面試官和面試者提供更準確的信息，從而提升面試的公平性和有效性。
Using AI for Online Real-Time Technical Interviews: Incentives for Interviewers and Interviewees
After watching related videos and online demos, here are my thoughts on using AI for online real-time technical interviews:
Main Evaluation Criteria for Interviews
Assuming future engineers will collaborate with other engineers, the main evaluation criteria for interviews include:
- Knowledge retention
- Programming proficiency
- Problem-solving approach
- Technical communication skills
- Behavioral questions
Limitations of Current Solutions
Current common online testing platforms largely meet the needs for evaluating knowledge retention and programming proficiency. In terms of assessing problem-solving approaches, these platforms can partially meet the needs since the way code is written can reveal thought processes. However, technical communication skills and behavioral questions cannot be effectively assessed through automated online testing platforms.
Importance of Technical Communication Skills
For technical communication skills, there are two main evaluation points:
- When the interviewee comes up with a good solution, they can make the interviewer understand it.
- When the interviewee encounters difficulties, the interviewer can provide hints that genuinely help the interviewee.
Meeting these two points indicates compatibility between the interviewer and the interviewee. To automate these evaluations, modeling the interviewer’s technical thoughts is required. However, if the AI model’s understanding of technical concepts significantly differs from the interviewer’s, it will not be suitable for assessing compatibility.
Corporate Image and Efficiency
Interviews also serve as a platform for corporate image promotion. Merely reducing the interview costs for the company might create an impression that the company does not value others’ time. If both parties turn on their cameras during the interview, it typically signifies trust. Therefore, if AI is to be used for online interviews, it should reduce costs for both the company and the job seeker, thereby attracting efficiency-oriented candidates to join the company.
Potential to Improve Interview Accuracy
The current discussion focuses mainly on how to use AI to mimic the interviewer’s process to enhance efficiency. However, I have reservations about the methods shown in online demos, as they do not increase the accuracy of finding the right candidates and might harm the corporate image.
The biggest problem in interviews is not spending too much time but accurately identifying who is suitable for the job. A common issue is that interviewees may fail to fully express themselves due to nervousness or unfamiliarity, thus missing out on opportunities. If AI can assist in guiding interviewees to provide more relevant information needed by the interviewer, it would help interviewers better understand the candidates.
Additionally, interviewers sometimes might be too hasty in their judgments, making decisions to reject candidates before fully understanding their thoughts. If AI can prompt interviewers to ask further questions or explore topics worth investigating, it could prevent overlooking potential talents.
In conclusion, I hope AI can enhance the accuracy of interviews, not just by pursuing efficiency but by assisting both interviewers and interviewees in providing more precise information, thereby improving the fairness and effectiveness of interviews.
