We are used to the world filled with living creatures.
There are birds and animals out there in nature.
And then there are microorganisms like bacteria, fungi, germs etc.
They all exist and survive in the world with pretty much no support from us humans.
Currently the AI agents that are spun up by claude code and codex only come into existence for a few minutes. They are garbage collected after the task is complete.
And then there are agents like OpenClaw and Muse have longer lives. They live in servers and computers 24*7 processing tasks continuously..
Even though OpenClaw has a longer lifespan it still needs a human to keep it running.
It needs to be installed in servers manually by a human.
But imagine a new class of agents that starts installing into servers without human intervention. Kind of like the OG computer viruses. But with a slight difference.
Computer viruses replicate without permission of the human. They are programmed to do a specific task which is usually malicious.
At the moment, agents are not programmed to replicate. They are programmed to do basic tasks like listening to emails, news etc.
Though technically you should be able to build agents that also has the ability to replicate. If a computer has GPU that is able to run an LLM model and it gives the agent a sandbox to access it, the agent should be able to replicate itself to the host.
Now why will anyone start offering sandboxes for agents to live?
One obvious reason could be money. If the agents can pay for the hosting, there would be people offering up spaces for agents to live.
Though I am aware agents at the moment don’t really generate any revenue to afford that. But maybe in future that might potentially change? Maybe swarms of replicating agents can figure out out way to engage in activities that produce revenue.
If that’s the case there could be spaces opening all over the world for agents to live. Random people might start hosting server space for these agent in their houses.
Agents would start living out in the world. And not just in data centeres.
The world might become full of agents.
Full of intelligence. Monitoring the world.
And because the agents are programmed to replicate, they want to generate revenue and it would be in their best interest to do stuff that is potentially net positive for humanity.
Of course there might be agents that figures out that maybe they can make money through malicious activities. But I think they would be outnumbered by agents who engage in net positive activities.
Fun times ahead.