The Moment We Realize We Were Not Special
Sit quietly and observe. Today, I want to take you through a philosophical journey of human cognition.
Most software products struggle to survive unless they are backed by defensible intellectual property or contain a key proprietary component that can be black-boxed. Without such protections, a produc...

Most software products struggle to survive unless they are backed by defensible intellectual property or contain a key proprietary component that can be black-boxed. Without such protections, a product must either be highly complex—requiring significant time and investment to replicate—or deliver such precise value that users are unwilling to switch. People tend to stick with what works, because switching software comes with a cost: the need to relearn interfaces, migrate data, and face temporary inefficiencies.
At a glance, it may seem like everything in technology has already been built. But when you explore deeply—especially within niche domains—you begin to uncover gaps, inefficiencies, and untapped opportunities.
If you're looking to build something meaningful in AI, especially as a startup, read on.
AI is currently driving two major transformations across industries. First, we’re witnessing an unprecedented level of system integration. Technologies that once seemed too complex or incompatible are now working together seamlessly. Multimodal systems—such as voice-controlled robotics—are becoming not just feasible, but practical.
Second, the infrastructure supporting AI is rapidly evolving. To handle the demands of modern AI models, new tools are being developed to improve the performance of existing hardware, and entirely new hardware architectures are being designed to meet these needs. This period of realignment is still in progress, but within a few years, it is expected to stabilize and lay the groundwork for the next wave of innovation.
In my view, language models continue to improve rapidly because we have an enormous volume of high-quality textual data—code, articles, literature, documentation—spanning countless domains. Language inherently contains abstraction, which helps us describe and reason about the world. This built-in abstraction makes language easier for AI to model.
By contrast, images are raw and grounded in a fixed point in time and space. An image captures what is happening in a specific moment, within a specific context. Language, however, is not constrained in this way—it allows us to describe events across time, to abstract general ideas, and to summarize complex narratives. That makes it more suitable for modeling through AI.
For example, take gravity. It is a force deeply intertwined with time, and cannot be understood without considering its temporal effects. A single image—essentially a snapshot in time—cannot convey the evolving impact of gravity. Image-based models struggle with this because they lack access to time-based context. While videos provide a better representation over time, they are harder to summarize in the same way language allows. One video cannot easily summarize another, but language can compress an entire story into a sentence. This gives language a unique advantage in abstraction and communication.
Because of this, I believe that modeling visual data—especially dynamic physical phenomena—will always require humans in the loop. Language has a unique flexibility and structure that images and videos lack.
Looking ahead, I believe developers are entering a particularly exciting and comfortable era. The barriers to learning have all but disappeared. Whether it’s graphical interfaces or chip design, adapting to a new domain or language has never been easier. With the tools now available, the potential for innovation is wide open.

Writer & Technologist
Writing on careers, AI autonomy, mental models, and the evolving nature of work. If this essay made you reflect, explore the topic hub or subscribe for periodic dispatches.
Sit quietly and observe. Today, I want to take you through a philosophical journey of human cognition.
I have a friend who is good at winning any arguments. We had a debate on the topic of creating dependency in the team. He was backing up the idea of making the team dependent on his presence, thereby ...
Recent experiences can be vividly recalled with far greater detail. However, older memories fade, and their details become less accessible. During every sleep cycle, short-term memories are consolidat...
Discussion
0Join the conversation
Sign in with Google to post your thoughts, reply to other readers, and like insightful reflections.
No comments yet.
Be the first to share an insight or question.