Keeping Up with Technology

Last Updated: 7/21/2025

"Any new and better way of doing things" is how Peter Thiel defines technology. As of June 11, 2025, it seems there is a new and better way of doing things every day. This is likely hyperbole, but in the field of software, the field I find myself in, it might as well be true. Every day, I am both deeply curious and excited about the advances while I also feel more strained to keep up.

Let's take humanity to have a “tech-tree” and extract a measure of “usefulness” from each technological advance. Then, we can chart usefulness against time. The graph we see would look nearly like a vertical line as you approach the present. And, as is the nature of exponentials, that vertical line will remain the frontier so long as humanity grows more branches on the tree.

For all of history, each new technology was a lever, extending our human brain and bodies to do things better. Today’s technology brings us to an inflection point. Instead of leveraging the brain and the body to do useful work, AI models, and their embodied forms, are the brain and the body and they do the work. Autonomy of that sort is categorically different from technology of the past.

In some workflows, humans have already been reduced to a “human in the loop” rather than a generative force. Now, I’m not here to give my opinions on AI ethics, the future of work, or any other such questions, nor am I here to give prescriptions of any kind. I simply want to unpack why I feel this strain to keep up with advances in autonomy. Perhaps, I can put it better by saying I feel perpetually behind learning about and using autonomy.

In reality, I’m not behind. But within myself, and within the circle of technologists and engineers, there is a pervasive, amorphous feeling of inadequacy to properly use and understand these tools.

There are two reasons for this.

The first reason is the technology is improving faster than we can discover the bounds of its application. I mean, autonomy is getting better faster than engineers can manifest the tooling to fully use the technology. To make this concrete, I will give an example from software engineering. Developers are creating all sorts of tools for language models. These tools include agentic programming, servers to give models context, browser use, and more. While improving, the rate of progress for tooling models have access to is slower than the rate of progress on the models themselves. This leaves the curious engineer constantly searching for ways to fully leverage autonomy which don’t yet exist or which they will need to create themselves.

The second reason is directly related to the qualitative difference between intelligent autonomy and previous technological advances. When we are automating uses for the human brain itself, there is less for the brain to figure out. Paradoxically, this means the better autonomy gets, the less need there is to keep up with autonomy. Plainly, instead of “trying to keep up” with AI like I would a previous technological advance like a new chip or device, I just ask AI how I can better use AI. The bottleneck is my own reasoning and intuition

For a large portion of use cases, there is no more keeping up. My feelings of strain or falling behind are a result of trying to bike on a treadmill. Technology is fundamentally different now, and my paradigm for interfacing technology needs to change with it. It's easier to get ahead than ever.