Understanding AI’s impact is tricky, especially when experts disagree. Some see heroic potential: Bill Gates thinks AI will help solve humanity’s greatest challenges like climate change, disease and education[1], and Google DeepMind co-founder Demis Hassabis calls it potentially “the most impactful technology” for advancing human flourishing[2]. Others warn of danger: Elon Musk calls AI “one of the biggest threats to humanity”[3], and UC Berkeley professor Stuart Russell suggests it could be “the last event in human history”[4]. Meanwhile, NYU professor Gary Marcus sees today’s AI systems as merely “party tricks” that are “hitting a wall”[5], while Peter Thiel points out how “we are running out of training data”[6].
This wide range of views reflects broader uncertainty. Surveys show most Americans aren’t too worried about AI taking their jobs[7], and when asked about the next 20 years, people are almost equally divided: about a third think AI will mostly help society, a third think it will mostly harm society, and a third aren’t sure[8].
With little consensus, there seems to be a feeling that we have time to figure it out. For example, around 60% of Americans believe robots will "never" surpass human intelligence, or that it's only likely "in the future". Only 15% think AI is "smarter than humans now." [9]
Yet reality is more nuanced: AI is already matching or exceeding human performance in several key areas. Human intelligence, far from being a monolith, is a collection of diverse capabilities provided by our brains. While there are aspects of human intelligence that AI hasn’t mastered — and some it may never master — it's steadily chipping away at slices of human intelligence, one by one, simultaneously.
More Advanced Than Most Realize
While we debate possible futures ranging from technological utopia to robot apocalypse, AI is already matching or surpassing human performance in several key areas. For example, research shows that AI systems have:
Outperformed average human university students in mathematics[9].
Nearly caught up to expert humans in other math and knowledge tests[10].
Surpassed non-experts in general knowledge assessments[11].
This contrasts with public perception, where only 15% believe AI is currently smarter than humans[12]. It would be more accurate to say that AI is already smarter than people in some ways, albeit still admittedly dumber than people in others. Yet unlike human brains, which are bound by the constraints of biology, AI doesn’t forget, it processes tons of data at high speeds, and it can be programmed to learn and improve nonstop, 24 hours a day. So perhaps the clearest-eyed view is that AI is already smarter than most humans in many ways, and it's systematically aspiring to become smarter than all humans in every way.
The disconnect between what AI can do and what most people realize it can do follows a pattern that Professor Clayton Christensen described in The Innovator’s Dilemma[13]: we often underestimate technologies that are reshaping our world. Our brains think in straight lines, but technologies like AI improve exponentially; starting with simple tasks before rapidly advancing to complex challenges.
We tend to swing between extremes: first overestimating how quickly a new technology will change everything, then becoming disappointed when it doesn’t match our wildest dreams. Ironically, it’s after these periods of disappointment (once the hype has blown over) when the most profound changes occur in earnest, often quietly and in the background.
Implications
There are tons of business and social implications—far too many to capture here. For example, businesses and nonprofits that wait until AI’s impact becomes more obvious will likely find themselves too late to catch up. The pace of technology improvement is simply faster than a lot of organizations' decision processes. Similarly, when it comes to social challenges, despite society’s equal division between AI optimists, pessimists, and the uncertain, most people appear unaware of the impact AI can already have on their work and daily lives. They’re in a race they neither see nor understand has already begun.
We’re at a critical juncture where AI capabilities aren't just approaching human levels but are already overlapping with them. This isn’t a future scenario—it’s happening now. This reality demands renewed urgency at three levels:
Personal: We need to get serious about updating our skills and rethinking our careers—not for some distant future, but for changes that are already happening.
Business: Organizations can’t afford to take a “wait and see” approach. They need to rebuild themselves now to work with AI, not scramble to catch up later.
Policy: We need practical rules and guidelines for bringing AI into society, not just theories and debates about what might happen someday. Policymakers also have to resist bouncing from one extreme to another, as public sentiment ping-pongs between overestimating and underestimating AI's impact.
While we debate AI’s future, it’s already reshaping our world regardless of how fast (or slow) people prefer to respond. At a minimum, the key to building a better tomorrow starts with more clearly understanding what AI can do right now—because that’s how we’ll harness its potential to solve problems and avoid dangers we once thought impossible.
Endnotes
[1]: Gates, B. The Age of AI Has Begun. GatesNotes, March 2024.
[2]: Hassabis, D. TED Interview, 2023.
[3]: Musk, E. Interview with Tucker Carlson, Fox News, April 2023.
[4]: Russell, S. Human Compatible: Artificial Intelligence and the Problem of Control. Viking, 2019.
[5]: Marcus, G. The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence. arXiv preprint, February 2024.
[6]: Thiel, P. Speech at Stanford Graduate School of Business, November 2023.
[7]: “Public Attitudes Toward AI and Automation.” Pew Research Center, April 2023.
[8]: “Global Views on Artificial Intelligence and Society.” Ipsos/World Economic Forum, January 2024.
[9]: “Artificial Intelligence (AI) Performance and Impact.” Our World in Data, 2024.
[10]: “AI vs. Human Performance Metrics: Comparative Studies.” Our World in Data, 2024.
[11]: “General Knowledge Assessment and AI’s Advancements.” Our World in Data, 2024.
[12]: “Views of Americans on Robot vs. Human Intelligence.” YouGov/Our World in Data, 2024.
[13]: Christensen, C. The Innovator’s Dilemma: When New Technologies Cause Great Firms to Fail. Harvard Business Review Press, 1997.