Agents are men-extenders
"At first I thought the real world was run by men, and then for one minute I thought it was run by horses, but now I realize that horses are just men-extenders." Ken, (Barbie 2023)
When Ryan Gosling's Ken delivers this line in Barbie, he stumbles onto something far deeper than the movie's satirical take on patriarchy. His observation about horses as "men-extenders" unintentionally captures one of the most transformative relationships in human history - and perhaps offers us a lens into our future.
For thousands of years, horses served as humanity's first true "agent" technology - semi-autonomous beings that could dramatically extend human capabilities while operating under human direction. When a human mounted a horse, they weren't simply gaining a mode of transportation; they were acquiring superhuman speed, strength, and endurance. A mounted human could suddenly cover vast distances at speeds that would have seemed magical to their ancestors, carry loads that would crush a person on foot, and achieve feats that reshaped the very fabric of civilization.
Just as horses amplified human physical capabilities, AI agents promise to extend our cognitive capabilities while working under our guidance. And just as the horse-human partnership reshaped society in ways that would have been impossible to predict, we could be close to a similar reshaping we need to prepare for.
I stumbled into this analogy while reading Razib Khan's excellent two part series on the history of the horse. And as we grapple with the our mental model about agents maybe we can look at the essentials of what they provide - they extend human cognitive capability.
About agents
There's a lot of debate about what exactly is an agent but I like the one in Chip Huyen's article on Agents. "An agent is anything that can perceive its environment and act upon that environment.
Artificial Intelligence: A Modern Approach (1995) defines an agent as anything that can be viewed as perceiving its environment through sensors and acting upon that environment through actuators".
And as work becomes less physical and more cognitive the ability of agents to sense and act becomes more important, extending our own human cognitive capabilities. And that is the frame within which we should evaluate what agents are and what they can do - focus at the cognitive tasks and workflows that you want to use AI to help you to perform.
What AI agents can do
One of the most insightful bits of information I have come across recently is not specifically about agents but more about a key building block for agents - LLM prompts.
This video by IndyDevDan classifies LLM prompts into six key areas and gives you a framework for deciding how to proceed with your business use cases.
- Expansion Prompts - generate content, explanation, ideas
- Compression Prompts - distills and summarizes information
- Conversion Prompts - transforms between formats
- Seeker Prompts - finds and retrieves specific information
- Action Prompts - execute commands and workflows
- Reasoning Prompts - provides judgements, conclusions, insights
More importantly it shifts the perspective from the agents to tasks. So if you are creating a GenAI strategy for your business you should organize your thoughts around the cognitive tasks your business perform, then figure out the block of cognitive computing you need to create to perform tasks and extend the capability of your workforce.
Agents and the future
Horses enabled modern society, allowing us to settle down, farm and create cities, and engage in long range trade and transport. Similarly by extending human cognitive capability agents and AI will bring significant positive impacts to society.
But lest we fall into unbridled positivity let's go back to the year 1219. In response to the seizure of a Mongol merchant caravan and the public execution of Mongol envoys sent in response, Genghis Khan unleashed an invasion of hundreds of thousands of horsemen into the heartland of the Khwarezmian Empire, razing the city of Bukhara to the ground until no trace was left.
Then, over the next few years the Mongols embarked on a 3500 mile arc of destruction throughout Eurasia dramatically reshaping empires. The rapidity of the change was enabled by the horse and the invention of the stirrup. We should be prepared.
About edgartools
edgartools is the most powerful way to navigate SEC filings in Python. It is also the easiest.
pip install edgartools
If you like it please leave a star on Github