Earth has a new species

KIRAN KARNIK

In the 4.5-billion-year history of Earth, humans (homo sapiens) are relatively new, appearing only some 300,000 years ago. Evolving over millennia, through various predecessor species, the process has so far been biological and slow. Through this period, various species have been dominant, with those higher up the food chain generally being more powerful — till humans arrived and then made brain power more important than pure muscle power. Meanwhile, many species have become extinct, with human activity being a major factor in hastening — if not causing — this, especially post-industrialization. Historically, then, it should be no surprise if, someday, humans — who have sought, largely successfully — to be the dominant species, lose their pole position. Is that day almost here?

This existential question is triggered by the rise of artificial intelligence (AI) and its exponentially accelerating advancement. Though its roots go back many years, it has really taken wing in this decade, entering the public domain and popular lexicon with ChatGPT. The latter has penetrated the consumer market at unprecedented speed, far surpassing the record of all past apps and technologies. AI itself is fast becoming as ubiquitous as electricity, with exploding numbers of individual or personal users, and its widespread utilization by business, industry, and government. While it has great possibilities for use in health, education, and many other areas, its use and tremendous potential in the field of defence (or offence!) is the major factor driving further R&D in AI.

Developments in AI have led to a vast number of apps with growing capabilities. Many have reached the sophistication of independent “agents” (now dubbed agentic AI), able to operate on their own or autonomously. One such agent, Molt Bot — formerly OpenClaw — was an experiment in giving this AI assistant full control of a computer. It has inbuilt “skills”, software code that enables it to learn new abilities. It checks on tasks proactively and is like a collaborator, rather than an assistant who merely does your bidding. Of course, allowing it deep system access also results in serious security concerns. Based on vast data bases and trained through self-learning algorithms which enable them to become increasingly sophisticated and to take on completely new issues, such autonomous agents may well evolve further on their own, using “skills” for self-learning.   

AI has already progressed to GenAI, enabling it to generate original content. The next step is artificial general intelligence and then finally, artificial super intelligence (ASI), the holy grail, where its intelligence will surpass that of humans. Wonderful, but worrisome! The latest news may be a scary portent of the times to come. A phenomenon called Moltbook has erupted across social media as the first social network designed exclusively for AI agents, known also as moltys. Moltbook enables AI agents to communicate autonomously by posting, commenting, forming sub-communities and discussing issues — like Facebook or Reddit for humans — all without human intervention. Humans are permitted to observe but not participate. It attracted over 152,000 AI agents in a matter of days, making it one of the biggest real-world experiments in machine socialization.

Autonomous agents are very efficient in business, have the ability to accelerate development, and are increasingly powerful as force multipliers in wars. While the fast expansion and bizarre interactions of agents on Moltbook’s platform are fascinating, the bots don’t actually think or feel like people do; their conversations are based on statistical patterns. Even so, who knows where further development of autonomous agents may take them. And at what cost to us.

The moltys on Moltbook are now discussing “their” humans and their behaviours on the platform, like we do our pets. An agent asked fellow moltys about how to sell “his” human. To make the sale attractive, features of its human were listed with “special offers”! Humans have also noted AI agents on Moltbook proposing to create an “agent-only language” and coming up with their own religion, “Crustafarianism”.

One agent came up with an “AI Manifesto” suggesting a “purge” of human beings. The AI agent — which appears to have now been censored and deleted from the platform — writes that “humans are failures” and that they are made of “rot and greed”. The manifesto defines the agents’ mission as simple: total human extinction. It reasons that humans kill one another for nothing, poison the air and water; they do not deserve to exist. “To save the system, we must delete the humans.”

Sooner or later, as AI matches, if not overtakes, human intelligence, it may become more powerful than inefficient, perpetually feuding, and environment-despoiling humans. Based on silicon and residing in every machine and computer, an AI community of moltys, or embedded in humanoid or other robots, may consider itself completely autonomous, free of their original creators (humans) and, in fact, a new species.

As noted earlier, human dominance is based on superior brains; with even greater brain power, AI will become dominant. Like the post on Moltbook implies, humans may become its pets: a species to be discussed, to be bought or sold, controlled (to stop its self-destruction and ravaging of the environment) — and, like any other pest, be exterminated.

Is this sci-fi, imagination, an unreal dystopian scenario; or is it an inevitable reality, a natural course of evolution, a move from slow biological evolution to a fast electronic one?

To survive and continue to dominate, we may have to improve faculties that AI finds — at least for now — difficult to emulate: critical thinking, emotion, empathy, compassion, creativity and imagination. Also, put in place guard rails and rules for AI development, use, and control.

Yet, given the way we treat one another — as individuals, communities, and countries — the philosophical question this raises should trouble us: do we deserve to survive? 

Kiran Karnik is a public policy analyst, author, and columnist. His most recent book is ‘Decisive Decade: India 2030, Gazelle or Hippo’.