Keeping It Human in a Digital World (3)

1 Heinrich von Kleist, “On the Gradual Construction of Thoughts During Speech,” trans. Michael Hamburger, German Life and Letters 5, no. 1, 1951, 42-46, https://www.ias-research.net/wp-content/uploads/2012/01/Kleist-and-Hamburger_-_1951_-_On-the-Gradual-Construction-of-Thoughts-During-Speech.pdf. 2 E. Xu, W. Wang, and Q. Wang, “The effectiveness of collaborative problem solving in promoting students’ critical thinking: A meta-analysis based on empirical literature,” Humanities and Social Sciences Communications 10, 16, 2023, https://doi.org/10.1057/s41599-023-01508-1. 3 Roger E. Beaty and Yoed N. Kenett, “Associative Thinking at the Core of Creativity,” Trends in Cognitive Sciences 27, no. 7, July 2023, 671-683, https://doi.org/10.1016/j.tics.2023.04.004. 4 Ethan Mollick, Co-Intelligence: Living and Working with AI, 2024, Portfolio/Penguin Random House. 5 Adam Cheyer, “What is the Case for Techno-Optimism Around AI?” Talk presented at Reinvent Futures, The AI Age Begins, 2024.

8. When Machines Act Human, What’s Left for Us?

Let me take you back to a watershed moment in March 2023—practically ancient history in AI’s rapid evolution. OpenAI had just revealed something stunning: their latest AI system at the time, GPT-4, wasn’t just passing human intelligence tests—it was absolutely crushing them.1 From achieving a 1410 on the SAT to acing Advanced Placement exams and even passing U.S. Bar, medical license, and certified sommelier tests, GPT-4 excelled at assessments that humans dedicate years to conquering, challenges we once believed were beyond the reach of machines.

When the report came out, I found myself staring for a long time at the page listing GPT-4’s high scores on thirty-four of our standardized tests, grappling with a swirl of emotions. It felt as though these were signposts pointing to a profound societal transformation. These tests have long been our yardsticks for human potential and achievement. While assessing machine “intelligence” using standardized tests has its limitations,[i] the results nonetheless prompt serious reflection. As the line blurs between our brightest minds and machines, we’re going to face some uncomfortable questions: What does this mean for us as individuals? For our education systems? For the organizations that have built their foundation on these metrics?

But it doesn’t stop there.

The Torrance Tests of Creative Thinking, a widely recognized assessment of creativity, placed GPT-4 in the top one percent for originality and fluency.2 A Wharton study showed that GPT-4 not only generates startup ideas much faster and more cheaply than humans, but the ideas are, on average, higher quality.3 For the first time, the creative abilities of AI, including the ability to generate original output, seem to match human abilities. And AI might persuade you more successfully than another person: research showed that AI is over eighty percent more likely to persuade you to the view it has been assigned than if you debate with a human.4 In a Google Research study, AI outperformed human physicians on twenty-four out of twenty-six scales measuring empathy and judgment.5 While of course AI can’t truly experience empathy—it has no actual emotions—this does show that it’s remarkably effective at simulating compassion. In fact, in some cases, AI is perceived as being more caring and understanding than humans themselves.6

One by one, machines are conquering skills we thought were exclusively ours. So where does this leave us?

The New Face of Competition: Redefining the Edge

We now face profound questions that few of us imagined we’d need to consider: What capabilities remain uniquely human? Where is our involvement not just helpful but essential? And perhaps most importantly, in which areas do we consciously choose human interaction over machine efficiency—even when AI might be equally capable?

As AI reshapes the way we work, learn, and live, we will need to understand how it alters the nature of competition and redefines success. Today’s college students will step into a world where their expensive degrees may no longer provide the advantage they once did—and they may need to demonstrate they can outperform AI or that they can leverage AI to be more capable than the technology by itself. Seasoned professionals, despite years of expertise, may find their value challenged as machines become capable of performing aspects of their roles. Whether we acknowledge it or not, humans are already in competition with AI (or people augmented with AI). While this may not have directly affected you yet, it’s likely to impact your life in the near future.

Over decades, we’ve made strides in understanding and enhancing our cognitive performance without AI’s assistance. We’ve gained insights into how sleep, nutrition, and environmental factors influence our brain function. We’ve refined our learning methodologies and continually evolved management science to optimize human and organizational performance.

This progress, though valuable, has been slow. And we may be approaching the upper limits of how far these methods can take us. AI offers us the opportunity to surpass these boundaries—to push beyond our natural limitations and, if used effectively, unlock cognitive feats that were previously unattainable. It’s quite possible the future will belong not just to those with the highest innate abilities, but to those who can most effectively collaborate with AI to amplify their human intelligence. I know this prospect can feel unsettling, but AI is not going away—and it will continue to advance.

iUsing standardized tests to evaluate AI systems presents significant challenges. These tests assess a narrow range of academic skills, failing to encompass the broader spectrum of human intelligence. AI systems, designed to process vast amounts of information quickly, have an inherent advantage on these tests, but this doesn’t necessarily reflect real-world problem-solving capabilities. Moreover, while AI excels at pattern recognition and information retrieval, it currently lacks the contextual understanding that humans bring to these tests, making direct comparisons between human and machine performance potentially misleading. Despite these limitations, AI’s performance on standardized tests serves as a directional signal of advancing capabilities and underscores the need for serious consideration of our role in a future where such systems are ubiquitous.