In this series, business journalist Timothy Goh provides practical answers to candid questions to address workplace challenges and advance your career. Sign up for more tips Straits Times Headstart Newsletter.
Generally, it is acceptable to use artificial intelligence tools such as ChatGPT for take-home interview assignments if your prospective employer allows it.
The more important question is how candidates use it. Michelle Koh, managing director at executive search firm The Edge Partnership, said it might be worth mentioning where and how AI was used to ensure transparency.
“Using AI to brainstorm ideas, structure thinking, improve the clarity of writing, and check grammar and sanity of formulas and code is becoming increasingly common and generally considered acceptable,” said Koh.
However, problems arise when candidates rely on AI to complete an entire assignment, especially when they are unable to explain or defend their research during the presentation stage.
“If you don’t really understand your output, your submission won’t reflect your actual ability,” she added.
Coe noted that even if a candidate ultimately secures the job, it may be difficult to retain it.
“I have personally tested AI-generated answers for financial reporting standards when hiring in the financial industry and found that the answers were sometimes inaccurate,” she said.
“Candidates who rely uncritically on such answers risk providing flawed analysis during the interview process.”
Kevin Chan, CEO of HR technology company Epitome Global, offered a different perspective.
He pointed out that take-home interview assignments are not intended to function like exams, but rather to simulate how candidates would work in real life, and this is increasingly including the use of AI tools.
“When I review the submissions I bring back, I’m not trying to verify whether a candidate can submit something completely unaided,” Chan says.
“What I care about is judgment: Did the candidate scope the problem well, ask smart questions, frame reasoning clearly, challenge weak assumptions, and produce something useful? Whether AI is involved is far more important than whether the results are good,” he added.
Chan said candidates who avoid AI entirely could also raise concerns for employers about whether they can adapt to evolving ways of working.
“In many roles today, refusing to use AI despite its obvious potential to help suggests that they may be struggling to adapt to the evolution of work,” he said.
Still, leveraging AI “well” is not the same as simply generating an answer and sending it directly.
The real skills tested are judgment and good sense. Can test takers recognize whether the AI’s initial answer is generic or weak, can spot factual errors or shallow reasoning, and can shape the output to reflect their own thinking, priorities, and communication style.
Chan said transparency remains important and candidates can ask interviewers upfront whether they are comfortable with the use of AI tools in assignments and whether they expect disclosures.
“Most recruiters will say yes, and many will quietly respect you for asking directly,” he said.
“If the answer is no, that’s still useful information. It gives you insight into how companies are thinking about technology and how adaptable it is.”
Mr Koh said many employers now expect candidates to use AI to some extent.
Others are interested in how candidates use it, like how proficiency with Google, Excel, and other productivity tools eventually became the norm.
That’s why candidates who use AI effectively while using original thinking are likely to outperform those who avoid AI altogether.
“A practical approach is to use AI as a co-pilot rather than a substitute…At the end of the day, you should be able to perform all parts of the assignment with confidence, clearly explain your reasoning, and ensure that your final submission is a true reflection of your abilities,” said Mr Koh.
