Most firms now use AI in HR processes, with human sign-off required 85% of the time: Survey

Large organisations are more likely than SMEs to use AI in HR functions, but on the whole employers still value human judgment, says the Singapore National Employers Federation.

An office worker typing on a laptop. (File photo: iStock)

This audio is generated by an AI tool.

SINGAPORE: Most firms in Singapore now use artificial intelligence for human resources processes, but humans still sign off on most decisions, according to a Singapore National Employers Federation (SNEF) survey released on Friday (Oct 30).

SNEF counted almost 3,000 use cases of AI in HR in its survey of 1,000 employers from June to July.

Out of this number, 85 per cent of use cases leave the final decision to human confirmation after AI provides input.

In the remaining 15 per cent, AI makes the decision without a human signing off on it. Among these use cases, 73 per cent are in “routine or high-volume administrative processes” such as resume screening and skills assessments, said SNEF.

CNA Games Guess Word Crack the word, one row at a time Buzzword Create words using the given letters Mini Sudoku Tiny puzzle, mighty brain teaser Mini Crossword Small grid, big challenge Word Search Spot as many words as you can Show More Show Less In nearly all use cases without human sign-off, employers still designate someone – such as the head of HR, supervisor or hiring manager – to intervene if needed and change or reverse the decision made by AI.

In nearly all use cases without human sign-off, employers still designate someone – such as the head of HR, supervisor or hiring manager – to intervene if needed and change or reverse the decision made by AI.

"Employers generally do not use AI in cases where there are more significant consequences for employees," said SNEF, adding that this reflects employers’ continued trust and emphasis on human judgment.

Large organisations are more likely to use AI in HR functions, with 81 per cent of firms with more than 200 employees doing so, compared to 65 per cent of small and medium enterprises (SMEs) with 200 or fewer employees.

SNEF president Kuah Boon Wee on Friday said that when AI is used in HR functions, the level of oversight should "match the stakes", with lighter monitoring for routine tasks and closer watching for decisions that affect people’s livelihoods.

"That is what responsible use should look like in practice," he said at the inaugural SNEF Employers’ Summit, with a theme of harnessing AI for business and workforce outcomes.

WHEN IS AI USED THE MOST?

SNEF’s survey also found that AI use in HR functions varies across an employee’s life cycle in an organisation.

AI adoption is heavier in the early stages of the employment relationship – for instance, to screen and shortlist candidates for recruitment, or to personalise employees' learning and inform their mobility within the company.

Even prior to hiring, AI is used in workforce planning, such as to forecast demand and identify skills gaps.

Companies also tap AI to analyse employee sentiment and attrition risk, said SNEF, as well as to benchmark compensation and analyse pay equity.

Reliance on AI is lowest towards the end of the employment relationship, when dealing with separation and exit matters.

At this stage, AI is used – if at all – mostly to analyse employee attrition data to reduce turnover and support separation planning or disciplinary investigations.

"Overall, employers are generally more likely to use AI in high-volume processes where AI can complement, rather than replace, professional judgement," said SNEF.

Said Mr Kuah: "Employers are already choosing carefully where to use AI, and that is the right instinct."

In the United States, makers of AI hiring tools Workday and Eightfold AI face lawsuits contesting how fairly these tools were used to screen candidates. These test cases are being watched for possible implications on AI use in HR processes.

Asked how concerns about potential bias would be addressed, SNEF chief executive Hao Shuo said employers are generally aware of the laws and their liability for offences of discrimination.

The survey found that 84 per cent of employers were very or extremely familiar with frameworks such as the Tripartite Guidelines on Fair Employment Practices, Fair Consideration Framework, Model AI Governance Framework for Generative AI, and the forthcoming Workplace Fairness Act, set to come into effect end-2027.

“When they deploy AI in the whole sifting process when CVs come in, employers will make sure that there is no bias in the process,” said Mr Hao.

He added that jobseekers can also learn how to make their resumes stand out in AI screening, and better understand how AI is deployed in the recruitment process.

Some employers already have safeguards in place, noted SNEF.

These include mechanisms to monitor the outcomes of AI tools and make adjustments such as reviewing decision patterns, retraining models, tracking reasons for human intervention and giving feedback to vendors.

FIRMS SEEK GUIDANCE ON AI

Employers’ top concern about using AI was that an over-reliance on the technology could erode HR capabilities over time, found the survey. This was cited by 39 per cent of respondents.

Roughly one-third were concerned about when to trust AI over professional judgment (36 per cent), accountability when AI makes errors (35 per cent), and lack of training to use the technology effectively (31 per cent).

For support in tackling such issues, 42 per cent of employers sought more clarity on their obligations and liability when using AI in HR functions, and training on responsible use of the technology.

Some also wanted an advisory service on using AI in HR, and a list of AI tools assessed for their fairness.

In response, SNEF has developed a "pre-scoped" solution for firms that integrates support for AI adoption with workforce transformation.

The solution covers four areas: strategic HR aligning workforce planning and employee development with business priorities; AI automation of routine HR work; workforce transformation involving job redesign and change management; and skills analytics using AI.

Firms can approach SNEF to be matched to specialist consultants who will help them transform their HR functions and workforce in a “people-centred manner”, said SNEF.

Five specialist consultants have been appointed: BDO, Elitez, EON, PwC and Skybots.

Eligible firms can fund these efforts through the Workforce Development Grant (Job Redesign+), which covers up to 70 per cent of costs for SMEs and 50 per cent for non-SMEs, capped at S$150,000.

Our chief editor shares analysis and picks of the week's biggest news every Saturday.

Get our pick of top stories and thought-provoking articles in your inbox

Stay updated with notifications for breaking news and our best stories

Join our channel for the top reads for the day on your preferred chat app

新加坡全国雇主联合会表示，大型企业比中小企业更有可能在人力资源职能中使用人工智能，但总体而言，雇主仍然重视人类的判断。

一名办公室职员正在笔记本电脑上打字。（图片来源：iStock）

这段音频是由人工智能工具生成的。

新加坡：新加坡全国雇主联合会（SNEF）周五（10月30日）发布的一项调查显示，新加坡大多数公司现在都使用人工智能进行人力资源流程，但大多数决策仍然需要人工签字。

SNEF 在 6 月至 7 月对 1000 位雇主进行的调查中，统计出近 3000 个 AI 在人力资源领域的应用案例。

在这些案例中，85% 的案例在 AI 提供输入后，最终决定仍由人工确认。

在剩余的15%案例中，人工智能无需人工审核即可做出决策。SNEF指出，在这些应用案例中，73%属于“常规或高容量的行政流程”，例如简历筛选和技能评估。

CNA 游戏 猜词游戏 逐行破解单词 流行词游戏 用给定的字母创建单词 迷你数独 小小的谜题，强大的脑力挑战 迷你填字游戏 小方格，大挑战 单词搜索 尽可能多地找出单词 显示更多 显示更少 在几乎所有无需人工签字的使用案例中，雇主仍然会指定某人（例如人力资源主管、主管或招聘经理）在必要时进行干预，并更改或撤销人工智能做出的决定。

在几乎所有无需人工签字的使用案例中，雇主仍然会指定某人（例如人力资源主管、主管或招聘经理）在必要时进行干预，并更改或撤销人工智能做出的决定。

SNEF表示：“如果人工智能会对员工造成更严重的后果，雇主通常不会使用人工智能。”他还补充说，这反映了雇主对人类判断的持续信任和重视。

大型组织更有可能在人力资源职能中使用人工智能，员工人数超过 200 人的公司中有 81% 这样做，而员工人数为 200 人或以下的中小型企业 (SME) 中只有 65% 这样做。

新加坡全国雇主联合会主席郭文伟周五表示，当人工智能应用于人力资源职能时，监督的程度应该“与风险相匹配”，对日常任务的监督力度要轻一些，而对影响人们生计的决策则要更加密切地关注。

“这才是负责任的使用在实践中应该有的样子，”他在首届SNEF雇主峰会上说道，该峰会的主题是利用人工智能实现商业和劳动力成果。

人工智能在哪些领域应用最为广泛？

SNEF 的调查还发现，人工智能在人力资源职能中的应用会随着员工在组织中的生命周期而变化。

在雇佣关系的早期阶段，人工智能的应用更为广泛——例如，用于筛选和甄选招聘候选人，或用于个性化员工学习和指导其在公司内部的流动。

甚至在招聘之前，人工智能就被用于劳动力规划，例如预测需求和识别技能差距。

SNEF表示，企业还会利用人工智能来分析员工情绪和流失风险，以及对标薪酬和分析薪酬公平性。

在雇佣关系的后期，处理离职和退出事宜时，对人工智能的依赖程度最低。

现阶段，人工智能（如果有的话）主要用于分析员工流失数据，以降低人员流动率，并支持离职计划或纪律调查。

“总体而言，雇主通常更倾向于在高容量流程中使用人工智能，因为人工智能可以补充而不是取代专业判断，”SNEF表示。

Kuah先生说：“雇主们已经在谨慎选择人工智能的应用领域，这是正确的做法。”

在美国，人工智能招聘工具Workday和Eightfold AI的开发商正面临诉讼，这些诉讼质疑这些工具在筛选候选人方面的公平性。这些案例正受到密切关注，以期对人工智能在人力资源流程中的应用产生潜在影响。

当被问及如何解决对潜在偏见的担忧时，SNEF首席执行官郝​​硕表示，雇主通常了解相关法律以及他们对歧视行为的责任。

调查发现，84% 的雇主非常熟悉或极其熟悉《公平就业实践三方准则》、《公平考量框架》、《生成式人工智能模型治理框架》以及即将于 2027 年底生效的《工作场所公平法案》等框架。

郝先生说：“当雇主在简历筛选的整个过程中部署人工智能时，就能确保过程中不存在任何偏见。”

他还补充说，求职者还可以学习如何让自己的简历在人工智能筛选中脱颖而出，并更好地了解人工智能在招聘过程中是如何应用的。

SNEF指出，一些雇主已经采取了保障措施。

这些机制包括监控人工智能工具的结果并进行调整，例如审查决策模式、重新训练模型、跟踪人工干预的原因以及向供应商提供反馈。

企业寻求人工智能方面的指导

调查发现，雇主对使用人工智能最主要的担忧是，过度依赖这项技术可能会随着时间的推移削弱人力资源能力。39%的受访者表达了这一担忧。

大约三分之一的人担心何时应该相信人工智能而不是专业判断（36%），人工智能出错时的责任（35%），以及缺乏有效使用该技术的培训（31%）。

为了获得解决此类问题的支持，42% 的雇主希望更清楚地了解他们在人力资源职能中使用人工智能时的义务和责任，以及负责任地使用该技术的培训。

有些人还希望获得有关在人力资源中使用人工智能的咨询服务，以及一份经过公平性评估的人工智能工具清单。

为此，SNEF 为企业开发了一种“预先规划”的解决方案，将人工智能应用支持与劳动力转型相结合。

该解决方案涵盖四个领域：战略性人力资源，将劳动力规划和员工发展与业务重点相结合；人工智能自动化日常人力资源工作；劳动力转型，包括工作重新设计和变革管理；以及使用人工智能进行技能分析。

SNEF表示，企业可以联系SNEF，由SNEF匹配专业顾问，帮助他们以“以人为本”的方式转变人力资源职能和员工队伍。

已聘请五家专业咨询公司：BDO、Elitez、EON、普华永道和Skybots。

符合条件的企业可以通过劳动力发展补助金（工作重新设计+）来资助这些工作，该补助金可覆盖中小企业高达 70% 的成本，非中小企业高达 50% 的成本，上限为 15 万新元。

我们的主编每周六都会分享对本周重大新闻的分析和精选。

订阅我们的邮件，即可获取精选热点新闻和引人深思的文章。

订阅通知，第一时间获取突发新闻和精彩报道。

加入我们的频道，即可在您常用的聊天应用上获取当日热门文章。