We must begin to protect the skills that make us human
The real value of AI was never in the outputs created, but in skills like stakeholder consultation, communication, and storytelling, learnt by the human doing the work.
As AI becomes embedded in everyday workflows, organisations will need to consider how to preserve human skills such as judgment, creativity and critical thinking. Image: Yeo Yong Kiat's LinkedIn Post
Seven months ago, I worked with US-based research lab Maiden Lab, focusing on emerging tech for public good and wrote a thinkpiece with Tim Marple about the role of governments in helping corporations re-design jobs in the age of artificial intelligence (AI).
Our thinking was centred around economic restructuring: shifting displaced workers from disappearing jobs into emerging jobs.
But the situation has now grown far more existential.
Re-designing jobs in the age of AI is no longer solely an economic restructuring exercise; it’s a critical strategy for preserving what makes us human.
Generative AI is replacing human cognition at scale
Generative AI (GenAI) is substituting human cognition at a rate that has grown alarmingly fast because of AI scaling laws, geopolitical competition and the vast human online farm that has gone by largely unregulated.
While it is easy to believe that humans will always have judgment, originality, and creativity, the concept of the "cognitive miser" challenges this narrative.
This term, introduced by pioneering American social psychologists Susan Fiske and Shelley Taylor, refers to humans’ inclination to think economically and to solve problems in the simplest way with the least possible cognitive effort.
Given a shortcut, like a reasonable AI-generated output, we will most definitely take it and not go back the long way around thinking for ourselves.
If such shortcuts are taken repeatedly, we will eventually lose our capacity for thought and judgement.
With advancements in AI, there is a myth that humans no longer need the skills to create things, such as corporate strategy decks, by themselves.
But the point is not what we create, but the skills we use and retain when we build things ourselves.
For example, stakeholder consultation, articulation, and storytelling are all skills built through the process of creation.
Some, in support of AI, tout its ability to break the inertia around starting a project.
We must question: when, and how, did plain thinking become a problem we needed to get around?
When individual judgment weakens, how much will the collective suffer?
The large language models (LLMs) underlying GenAI are just next-word prediction machines, which means that all the outputs are merely statistical averages.
When everyone starts to use the same AI tools trained on the same datasets, we must expect that outputs will converge eventually, and people will derive similar answers over time.
We must also expect that there will be less impetus for people to consult, challenge, or contribute to someone else's perspective.
There will be faster convergence on set solutions, at the price of fewer constructive debates and less diversity of thought.
The individual phenomenon of the cognitive miser would rapidly scale up to a sociological one.
What we stand to lose is not just judgment or the ability to create, but two types of human skills that make us human in the collective.
The first type of human skills is social-emotional: motivation, self-awareness, curiosity, lifelong learning, leadership, and empathy.
But when the frequency of constructive debates falls off, we risk losing the ability to hold our own opinions in the face of a sparring partner.
The pure joy of learning is a very fragile force. When we feel there are no more missing pieces to our knowledge, it is hard to start questioning because we no longer want something we do not have.
The thrill of deriving a conclusion will be no more. We will shift from curious learning to simply knowing.
The second type of human skills we stand to lose is judgment itself.
But what is judgment? To simply say that it is the ability to analyse a scenario, minimises that which we are potentially ceding to our machine masters.
For example, attention to detail is what we train when we build something.
Or when we create something for the human gaze, we hone our skills at storytelling, both implicit and explicit.
Understanding and framing a problem are fundamental to creation, especially in a knowledge-based economy.
We give up the opportunity to develop it when we give up the act of doing or thinking.
We also give up many other things in this class of skills, such as systems thinking, or causal analysis.
Judgment is not just a singular skill, but a combination of various skills, consolidated through our experience of doing and thinking.
Protect our skills through structures and processes
It might seem that I'm anti-AI, but that's not what I'm saying.
I'm saying that just as we are all adopting it, we need to care about how we protect what's left of what makes us human.
This cannot be done through the romantic notion that everyone will use AI wisely.
The solution is to create structures and processes that protect what we value.
There is the fear that as AI absorbs entry-level tasks, organisations will expect new hires to perform without the traditional warm-up period that helps them build competency and clarifies expectations for them.
It is likely we will then see dwindling numbers in senior leadership, and a lack of original thought in those who remain.
One way to prevent this is to mandate the warm-up period in HR processes, or to delineate the core skills that require protection from AI usage.
AI is making its way through our economy and organisations at a breakneck speed, not just by the pace of technological change, but also by how organisations have embedded the use of AI throughout our workflows.
Thus, the natural solution is also through workflow, by embedding protective policies into them.
And we should start now just as we start the race for AI adoption.
------------------
Yeo Yong Kiat is the Head of Artificial Intelligence at the Health Sciences Authority of Singapore.
The original article was first published on Yeo's LinkedIn page here, and then edited.