As AI Gets Better, We Have to Get Better at Thinking

I keep hearing some version of the same prediction about the future of work: AI will take on more of the routine work and free people up to focus on higher-value work. I believe that, but lately I’ve been thinking about the assumption buried inside it. What makes us think people will automatically be ready to do the higher-value work?

If AI handles more of the analysis, research, drafting and repetitive problem-solving that consumes our time today, the work that remains for people will increasingly depend on creativity, critical thinking and judgment. Yet I hear far more discussion about teaching people how to use AI than I do about developing those capabilities. We’re investing in AI literacy, new tools and adoption, all of which is necessary. But if the promise is that AI will free people up for higher-value work, we should be investing just as deliberately in the higher-value skills we expect them to use.

Higher-value work requires higher-value skills

I don’t think we can assume those skills will simply emerge. People can become better at framing problems, examining assumptions, making connections across disciplines, considering different perspectives and exercising judgment. Creativity and critical thinking aren’t fixed traits that some employees happen to possess. They are capabilities that can be developed. The question is whether companies will put the same energy into developing them that they are putting into teaching people how to use the technology.

There’s another issue here that I think we’re only beginning to appreciate. Some of the work AI is starting to take on is the same work through which people have traditionally developed judgment. You analyze enough forecasts and you begin to recognize when a number doesn’t make sense. You wrestle with enough difficult problems and learn to approach them from different angles. You make decisions, get some wrong and develop better instincts.

That raises an important question. If AI increasingly does the research, builds the first analysis and develops the initial recommendation, how will people develop the experience they need to exercise judgment over that work later? We could find ourselves removing some of the very work through which people learn to do the higher-value work we expect them to move into.

The better AI gets, the more judgment matters

There’s a second challenge. AI itself can make these human capabilities easier to neglect. A system that's wrong half the time gets checked constantly. A system that's right most of the time earns our confidence, and eventually we stop looking quite as closely.

Imagine an AI app that's produced a useful forecast month after month. Its analysis has consistently been solid and its recommendations generally make sense. At some point, the question changes from “Is this right?” to “Is there any reason to think this is wrong?” That may sound like a subtle distinction, but it changes the role AI plays in the decision. Instead of being another source of input to evaluate, its recommendation can gradually become the starting point around which everyone else thinks.

That’s where automation bias and overreliance become real business issues. The concern isn’t simply that AI will occasionally produce a bad answer. We may give that answer more weight than it deserves because of all the good answers that came before it. We don’t need people recreating every calculation AI performs. We need people who recognize when something deserves a second look. An assumption doesn’t seem right. An outlier was dismissed too quickly. A recommendation makes sense mathematically but doesn’t square with what we know about the customer or the market. As AI gets better at producing answers, human judgment increasingly means knowing when to keep asking questions.

People have to be able to challenge the answer

There’s one more piece. You can develop great critical thinkers and still get very little value from them if the culture teaches them to stay quiet. People figure out quickly what happens when they question a recommendation, challenge the prevailing view or tell an executive they see something differently. If thoughtful pushback is dismissed often enough, people stop offering it.

That’s why trust and psychological safety belong in the AI conversation. We need people who are comfortable challenging an AI-generated recommendation, but we also need them comfortable challenging their manager, the CFO or the CEO. The real test is whether someone who sees something that doesn’t look right will say so before it becomes a decision.

Companies are spending enormous amounts of time and money preparing their technology for the future of work. We should be asking whether we’re putting the same effort into preparing people for it.

If higher-value human work is part of the promise of AI, developing higher-value human capability has to become part of the AI strategy. We need to think about how people develop those capabilities, how they maintain their judgment as AI becomes more capable, and whether our cultures actually encourage them to use it.

As AI gets better, we need better at thinking.  That only comes from intentionally building capability.

About Root Idea

Root Idea helps CFOs protect the business from AI decision risk. We work directly alongside finance teams to map AI influence, establishes decision governance controls that hold up to board scrutiny, and delivers training and change management to make governance stick.

If your organization is scaling AI and governance hasn't kept pace that's exactly the conversation we're built for. Learn more at rootidea.ai.