We’re all living through a rare pivotal point in history, where the ground beneath our feet is shifting faster than most people can process. The rate of change seems to be accelerating, and while humans are remarkably adaptable, we’re also notoriously bad at predicting the downstream effects of new technologies.
I think it’s a safe bet that kids born today will never be smarter than AI. By the time they’re old enough to ask, “What was life like before AI?” the question will feel as strange and abstract as asking a grandparent what life was like before telephones.
OpenAI released GPT-5 today, the newest version of their ChatGPT and most advanced large-scale AI model. The company has made GPT-5 available to everyone, including its free users.
The new GPT-5 interface can take on hard scientific challenges, answer technical questions that require mastery and expertise, spin up software from scratch, and write with a tone and clarity that feels noticeably more natural than previous GPT-4 models. It also switches automatically to select the appropriate AI engine, and then it just does stuff as Ethan Mollick explains in his recent article.
“GPT-5 is not one model as much as it is a switch that selects among multiple GPT-5 models of various sizes and abilities. When you ask GPT-5 for something, the AI decides which model to use and how much effort to put into “thinking.” It just does it for you. For most people, this automation will be helpful, and the results might even be shocking, because, having only used default older models, they will get to see what a Reasoner can accomplish on hard problems. But for people who use AI more seriously, there is an issue: GPT-5 is somewhat arbitrary about deciding what a hard problem is.”
– Ethan Mollick, Wharton School of the University of Pennsylvania
Adoption of LLMs and this new breed of AI has been sudden and widespread.
The chart above shows the amount of time it takes to reach 1 billion active users. In early 2025, according to Coatue EMW research, the user base of ChatGPT doubled in a matter of months.
New frontier models are brilliant and quick at daily work that takes seconds or minutes, but still developing endurance for long horizon tasks that take experts thousands of hours. But that’s likely to change soon.
What Might Change Next
Many AI experts agree that within 2-3 years, new frontier AI models will be directly responsible for a major scientific breakthrough. It won’t be a matter of whether AI can answer your question accurately anymore. The most strategic questions that people ask might be the ones that actually matter.
Even as new emerging innovative systems go online, solving novel problems might need more than just analyzing data we already have today with more massive processing power. New breakthroughs might require tools and instruments purpose-built for experimentation at AI scale. I’m convinced there’s a large capability overhang, where last year’s AI models we’ve all been using have millions of use cases that haven’t been approached or explored yet.
Exploring the possibilities of emerging systems will likely be uneven. Recently, Anthropic’s CEO Dario Amide said half of all entry-level jobs will be eliminated. Amide believes AI will cut half of all entry-level white-collar jobs within five years, a move that could cause unemployment to spike significantly.
Many worry about the opportunities for new graduates, however I believe that a 22-year-old can rapidly adapt to new digital tech, while an average 62-year-old may not. As middle management becomes more responsible for balancing the performance of both humans and AI agents, seasoned knowledge workers with decades of experience might want to seriously consider reskilling in the age of AI.
There’s never been a better moment to learn a new skill, launch a startup or invent something new. The AI tools in your pocket today give you access to processes which required departments of hundreds of people to accomplish in previous decades.
Meanwhile, the boundary of what we consider “real” is sliding. Have you seen the recent bunnies on the trampoline video?
The video above is interesting because it’s one of the first videos that went viral before people generally realized it was generated by AI.
Several friends of mine are rightfully concerned about rise of AI media, and our inability to distinguish between real or generative media. Some artists and producers are hesitant to leverage new AI tools, while others are afraid to admit they’re already using them for inspiration and production. If you scroll through photos on your smartphone today, they’re all already processed through layers of algorithmic interpretation before you ever see them. This threshold of authenticity could be a moving target.
Open Questions
OpenAI’s Sam Altman seems to be grappling with big questions in recent interviews. Some of these questions sound like science fiction but might be closer than they seem: What happens if GPT-6 discovers new science? Could new frontier models cure diseases? What if new AI tools also create biosecurity risks? Could a future war be fought over access to AGI compute? How long until AIs generate more words in a single day than every human on Earth combined that day, and what if an AI researcher changes parameters of a widely adopted model’s personality in the wrong direction? Ripple effects of AI now have global cultural impact.
So what can you and I do? We get fluent in AI. Stop waiting for the “right moment” and just experiment now. Figure out how these incredible new tools fit into your work and your life before rules solidify. Major technological shifts create incredible possibilities.