- Reward Management
- Salary Surveys
- Reward Consulting
- Performance & Development
- Job Design & Job Grading
Tools such as ChatGPT, Copilot and Gemini are increasingly taking over tasks, from text and image production to programming, data analysis and decision support. With AI now capable of doing more and more, the question arises as to whether human reasoning skills are still necessary. Because AI doesn't really think. You do.
The rise of generative AI has fundamentally changed the way we work. Tools such as ChatGPT, Copilot and Gemini are increasingly taking over tasks, from text and image production to programming, data analysis and decision support.
With AI now capable of doing more and more, the question arises as to whether human reasoning skills are still necessary.
After all, if AI is already doing “all the 'thinking,' what value can human reasoning add?".
At Hudson, we firmly believe in the added value of human reasoning. In fact, cognitive skills like verbal, abstract and numerical reasoning are more relevant today than ever. In this blog, we will explain why and how measuring classic cognitive skills continues to play a crucial role in selecting 'AI-ready' talent.
While AI can generate, structure and present information at lightning speed, it lacks something fundamental: true understanding. AI output is based on patterns and statistics, not on context, ethics or experience. Most users know by now that AI output can sound very convincing but at the same time contain faulty reasoning and conclusions. Therefore, human assessment (a "human in the loop") is essential to interpret, evaluate and correct AI output. This requires cognitive skills that AI itself does not possess.
AI tools can generate impressive texts, but they are not flawless. The generated texts can sometimes be illogical and contain bias or even outright hallucinations. Verbal reasoning enables AI users to recognise and correct these errors.
AI recognises patterns but lacks the ability to create new concepts or to make unexpected connections. Abstract reasoning enables people to integrate AI output into broader frameworks, to think strategically and to solve complex problems. It also helps when dealing with completely new or ambiguous types of information and problems. This is essential in strategic roles: AI can provide input, but it is the person who uses that input who understands the full context, sees the links with other domains and ultimately makes the right decisions.
While AI can analyse data, it does not always interpret it correctly. For example, an AI dashboard may be showing a rising line, but is this increase really meaningful? Could the relationship be causal or not? In these types of situations, numerical reasoning makes the difference between working on blind trust and taking informed action.
A recent article in De Standaard states that while AI makes some tasks more accessible to less experienced employees, it is primarily the more highly educated professionals who can make the smartest use of AI. This will lead to less executive support being needed in teams, which means that jobs are likely disappear, thereby triggering an increase in demand for upskilling and reskilling. One thing is certain: those who are cognitively strong will remain relevant. Those who can efficiently manage, assess, correct, and strategically deploy AI will become more valuable to employers. Therefore, cognitive skills have not become redundant because of AI; they are a lever for sustainable employability.
We are sometimes asked about tests for evaluating AI skills. This is understandable: AI is ubiquitous, and it would seem logical to want to measure how well job candidates can work with it. Nevertheless, it is important to consider the limitations of any specific AI skills test.
Developing a reliable and valid (selection) test requires extensive research, a process that can easily take 12 months or more. By the time an AI skills test is ready for use in ‘high-stake’ procedures, the technology on which it is based may already have changed.
What is currently considered a relevant AI skill may in a few months be outdated. The AI landscape is evolving rapidly, and the tools that are dominant now can be replaced by new technologies tomorrow. Therefore, a test developed today may already be obsolete by the time of its publication.
What remains stable are the underlying cognitive skills needed to work with any AI tool, i.e. critical thinking, inductive reasoning, abstraction and evaluation. These skills are not tied to one specific technology and are therefore futureproof. In short: even if AI technologies and their corresponding interfaces change, the evaluative and critical skills needed to work with them remain largely the same. Classic reasoning tests measure exactly those skills and are scientifically sound, broadly applicable and sustainable in a rapidly changing world.
Cognitive skills form the basis for AI literacy. Therefore, psychometric reasoning tests remain an essential instrument in high-stake selections. Would you like to know more about how to select candidates who are ready for the AI future? We will be happy to help you.
Submit your HR challenge to us. Together we look at how we can help you.