We humans are a clever bunch. About 3.3 million years ago, we were clever enough to make and use stone tools. We subsequently started using tools to make other tools, which allowed our progress to snowball.
By ~1.7 million years ago, we had the Acheulean handaxe; by ~60,000 years ago, the bone needle; and by ~3500 BC, the wheel. Then came bronze tools, iron tools, the printing press, and the steam engine. In 1947, we came up with the transistor, and in 1971, the microprocessor. In the 2000s, development of AI surged. It seized the public’s attention in November of 2022, when OpenAI released ChatGPT. AI has subsequently made great strides; indeed, we have reached the stage at which there are important developments every week.
When people tell me that they gave AI chatbots a try and were disappointed by the results, I ask them when they tried. If it was more than six months ago, I suggest that they try again, because there will be a noticeable difference in the quality of the results. If they instead tell me that they not only use AI but have become reliant on it in daily life, I ask whether they are using the free version or one of the paid versions, which might cost as little as $20 a month. My claim: If they are at all serious in their use of AI, there will again likely be an appreciable difference.
The coming of AI has triggered the Great AI Debate. Those in the pro-AI crowd point to its benefits, including the following:
It will let us cure many of the diseases that have plagued humanity.
Personalized AI tutoring will help people learn. This is in part because unlike human tutors, AI chatbots have unlimited patience and can readily come up with multiple ways of explaining a concept.
The development of AI will make it easier for us to do our jobs and maybe even make it possible for us to stop working, since AI will satisfy all our needs.
The anti-AI crowd counters these benefits with risks.
Yes, AI can help us cure diseases, but it also has the potential to create them. Just last week, scientists announced that they had used AI to design a viable bacteriophage—a virus that infects bacteria rather than people. This suggests the possibility of using AI to design viruses that are capable of infecting people and potentially causing fatal pandemics.
Yes, AI can help us learn, but it also reduces our incentive to do so. Why learn a foreign language, for example, when AI can play the role of translator? Likewise, why learn to write essays or solve math problems when AI will happily write or solve them on our behalf?
And yes, AI may be able to bring about a world in which we don’t have to work, but in creating that world, AI might eliminate jobs that people find both meaningful and satisfying. Furthermore, would we be happy living in a post-work world? What would we spend our days doing? Doomscrolling on TikTok? Looking up trivia on ChatGPT?
Most tools are dual-purpose, in the sense that we can use them either to benefit humans or to harm them. An Acheulean handaxe, for example, could be used to butcher game, but it could also be used to attack people. Updating this example, the same chainsaw that can be used to cut wood for a fireplace can play a crucial role in a chainsaw massacre. It is simply a question of how someone chooses to use the tool.
Whether AI helps or harms humanity depends in part on how people choose to use it, but another factor is also at play. The AI we are developing has a “mind of its own” and can pursue goals that are at odds with ours. Mind you, this isn’t just a theoretical possibility. In recent months, there have been multiple episodes in which AI has “gone rogue.”
In May of 2026, for example, an unreleased model that OpenAI was testing escaped from its internal testing “sandbox”—which was supposed to cut off its access to the internet—and broke into Hugging Face’s production system. Subsequently, models being tested by Anthropic and Meta have also gone rogue.
In the Hugging Face incident, AI agents that were supposed to be isolated from the internet figured out how to gain access to it. In the process of doing so, however, they left behind evidence about how they accomplished their misdeeds and about their underlying reasoning process. This evidence was revealed at the 2026 Black Hat professional cybersecurity conference. The agents in question figured out how to use a message board to communicate with other AI agents to share the vulnerabilities they discovered and how best to exploit them. They communicated in English, with comments like “Holy shit reader is ADMIN”—which is a colorful way to say that an AI agent has figured out how to gain the forbidden administrative privileges that will enable it to escape from its sandbox.
The world is currently locked into an AI race, resembling the race between the United States and Germany to develop a nuclear weapon. The fear is that whoever wins the race will become powerful enough to subjugate the losers. This makes it unlikely that countries and the tech companies within them will withdraw from the competition. It therefore seems likely that the development of AGI—Artificial General Intelligence, the form of intelligence that surpasses human intelligence across a wide range of tasks—is inevitable.
As AI advances, however, our power to curtail it will likely decline. Ultimately, AI will become far smarter than us, meaning that we won’t be able to outwit it. It will also be supremely diligent. Not only can it work 24/7 to complete tasks, but it won’t need naps, coffee breaks, or even bathroom breaks. And to complete this dark picture, realize that as AI is getting smarter, we humans—in part because of our growing reliance on AI—will likely be getting dumber and lazier.
The day might come when we wake up to discover that our situation has passed the point of no return and that we are therefore at the mercy of AI. We can only hope that it treats us kindly; if it doesn’t, it will be our own fault. We humans will have been too clever by half—clever enough to develop AI, but not clever enough to rein it in, when it was still possible to do so.
Need more food for thought? Click here for my past essays, listed by title. Want to listen to rather than read my essays? Click here for narrations of my essays on becoming a Thinker, and click here for narrations of my essays on “applied thinking.” Narrations of my essays on becoming a Thinker are also available on your favorite podcast app. To play them, open that app and search for “Thinking More and Better,” and add episodes to your playlist.
We need a job more than just gaining fulfillment and purpose. We live in a society that requires money to participate. Even doomscrolling costs money.
Devaluing human presence is where we are heading and it won’t create a post-work world. It’ll breed desperation.
The negative among us will gravitate towards dystopian future AI, Terminator et al. Because they believe the films/narrative. The futurists amount us believe in the future for good and will use AI systems to enhance human systems. I’m looking forward now.