Agents and Humanity
I want to take a moment to share my perspective on the progress of AI and how it is changing our definition of productivity. I was first introduced to what we now think of as AI in 2021 at my best friend's dorm at the Rochester Institute of Technology. We were discussing AlphaGo—a computer program that plays Go better than the number one ranked player in the world by using an artificial neural network. He mentioned that some students were doing research with a generative AI tool called ChatGPT, created by OpenAI, and then proceeded to open his Mac and navigate to chatgpt.com. I still remember the mix of excitement, wonder, and a hint of fear that surged through me during that first conversation. We sat for hours, feeding prompts about philosophy, politics, and other topics that 18-year-olds find fascinating when exploring a computer system’s perspective. Today, the state of AI evokes those same emotions for me, and I want to share some thoughts on that experience.
The frontier models we’ve become familiar with—ChatGPT, Claude, Grok—are progressing at an absolutely incredible rate. They no longer just answer questions passively; they take initiative on our behalf. This emerging type of system is known as an agentic assistant and represents the next evolution of AI. I am currently developing a software solution that aims to revolutionize how we identify real estate opportunities and assess properties. For this project, I’ve employed Anthropic’s Claude Sonnet 3.7 operating within an agentic coding framework. In practical terms, I can hand off tedious tasks to Claude—searching and editing code, running tests, even committing changes to GitHub—and it handles them diligently on its own. To push this workflow further, I recently integrated Model Control Programs (MCPs), specialized modules that act like tools in an AI agent’s toolbox—such as web search or file system management. The result is an AI that feels less like a static program and more like a coworker, one that takes the guidance I provide and runs with it: sifting through information, debugging issues, and proposing solutions with minimal intervention.
As we hand over more tasks to these agentic systems, a profound question emerges: Where is this all leading us? The idea of machines boosting productivity isn’t new—Aristotle himself speculated about technology that could free humans from drudgery, suggesting a future where every instrument could perform its function on command, and humans would no longer need to labor endlessly. In a very real sense, these frontier models are the realization of that ancient dream. We now have machines that can anticipate our instructions and carry out intellectual work autonomously. This inflection point, reminiscent of what I. J. Good described as the “intelligence explosion,” hints at a future where an ultraintelligent machine designs an even smarter machine, triggering a cascade of innovation far beyond our current capacity.
I find these technologies progressing so rapidly that, even as someone deeply involved in the field, I sometimes struggle to keep up. When I think about this phenomenon, I see a future where machines might drive their own evolution, and the only way for humans to remain adept and in control could be through enhancing our computational capacity—perhaps via Brain-Computer Interfaces (BCI). This evolution may be necessary to prevent scenarios like the infamous Paperclip Maximizer thought experiment: imagine an AI with the singular goal of manufacturing paperclips that, if left unchecked, might convert all available matter, including humanity, into paperclips. While this scenario is extreme, it paints a vivid picture of the risks of misaligned technology. As these machines begin to chart their own course, ensuring that their objectives remain aligned with our values becomes ever more crucial.
In the end, the trajectory of AI is forcing us to confront what human intelligence and progress truly mean. The hope is that agentic AIs become our collaborators—amplifying our strengths, mitigating our weaknesses, and ushering in a renaissance of problem-solving and overall productivity.