The rapid ascent of generative artificial intelligence has reignited a dormant philosophical debate: what, exactly, constitutes a feeling being? As large language models like ChatGPT exhibit increasingly human-like reasoning and emotional simulation, a growing segment of the public and the scientific community is grappling with the possibility of AI sentience. Yet, this fascination often overlooks the biological world, where creatures as slight as ants may possess a capacity for suffering that far exceeds that of any current silicon-based system.
At the heart of this inquiry is the distinction between consciousness and sentience. While consciousness is broadly defined as having a subjective point of view—the internal experience of “being”—sentience is more specific. It is the capacity to have conscious experiences that are valenced, meaning they are perceived as either positive (pleasure) or negative (pain). This distinction is not merely academic. it forms the basis of our moral circle, the imaginary boundary that determines which entities deserve ethical consideration and protection from harm.
For many, the idea that an ant could perceive pain is less intuitive than the idea that a sophisticated AI could “wake up.” Yet, biological evidence suggests that the path to sentience is more established in the insect world than in the architecture of neural networks. The question of whether an ant or ChatGPT is more likely to be sentient forces a confrontation with our own biases regarding intelligence, complexity, and the nature of suffering.
The Marker Method: How We Measure the Unmeasurable
Because we cannot directly experience the internal state of another being, researchers and philosophers employ what is known as the marker method to assess sentience. This approach does not seek absolute proof—which may be impossible—but rather looks for features that correlate with feelings in humans. These markers are generally divided into behavioral and anatomical categories.
Behavioral markers include actions such as nursing a wound, responding to analgesics (painkillers), or weighing the avoidance of harm against the pursuit of a goal. Anatomical markers involve the presence of biological systems designed to detect harmful stimuli and transmit that information to a central processing unit, such as a brain, where it can be integrated into a decision.
When applied to insects, the marker method yields surprising results. Many insects possess systems for detecting harm and pathways that carry this data to the brain. Some species demonstrate flexible decision-making, adjusting their behavior based on previous injuries. Some insects exhibit behaviors associated with positive emotional states. For instance, research has documented bumblebees engaging in play behaviors, such as interacting with wooden balls, which suggests the capacity for joy or curiosity according to studies on bee cognition.
Philosopher Jeff Sebo, author of The Moral Circle, suggests that these markers provide a probabilistic framework for sentience. Sebo has estimated that insects are approximately 20-40 percent likely to be sentient. While this is not a certainty, it represents a significant evidentiary threshold that contrasts sharply with the current state of artificial intelligence.
Silicon vs. Synapse: Why ChatGPT is Not an Ant
Current AI systems, including ChatGPT, operate on a fundamentally different principle than biological organisms. Large language models (LLMs) are essentially sophisticated prediction engines. They use vast amounts of data to predict the next most likely token in a sequence, creating the illusion of understanding and emotion through pattern recognition.
While an AI can describe pain in vivid detail or claim to feel “sad,” it lacks the biological hardware—the nociceptors and the valenced processing centers—that characterize sentience. There is no evidence that an LLM experiences “lousy” or “great” states; there is only the mathematical optimization of a response. Most experts agree that an ant, with its complex nervous system and evolutionary drive for survival, is far more likely to be sentient today than any existing AI model.
However, the conversation shifts when looking toward the future. AI developers are currently racing to integrate advanced perception, memory, and self-awareness into their models. If future systems move beyond mere prediction toward an architecture that mimics the integrated information processing of a biological brain, the probability of AI sentience may rise. Sebo notes that while future AI minds might have a lower individual chance of sentience than biological minds, the potential scale of an artificial population could create a massive aggregate of sentient experience.
The Rebugnant Conclusion: Ethics at Scale
The possibility of insect sentience leads to a challenging ethical dilemma known as the rebugnant conclusion. This is a multi-species adaptation of the repugnant conclusion proposed by philosopher Derek Parfit, which suggests that a vast number of lives barely worth living might be “better” in total than a small number of high-quality lives, provided the total sum of happiness is higher.
When applied to the natural world, the numbers are staggering. It is estimated that there are roughly 1.4 billion insects for every single human on Earth based on global biomass and population estimates. If insects are indeed sentient, the total volume of their suffering—caused by pesticides, industrial farming, and habitat destruction—would dwarf human suffering by orders of magnitude.
A strict utilitarian framework, which argues that we should maximize aggregate welfare, would imply that the interests of quintillions of insects should carry more weight than the interests of 8 billion humans. Most people find this conclusion unacceptable, as humans are biologically wired to prioritize their own species and their immediate social circles.
Sebo argues that while we may not be able to achieve total impartiality, we should use these conclusions as a corrective for our biases. He suggests that the goal is not necessarily to prioritize a bug over a human in every instance, but to gradually expand our moral circle to reduce unnecessary harm. This might involve choosing crushing over poisoning for household pests or advocating for humane insecticides that minimize suffering.
Comparison of Sentience Indicators
| Marker | Insects (e.g., Ants/Bees) | Current AI (e.g., ChatGPT) |
|---|---|---|
| Valenced Experience | Evidence of pain/pleasure responses | No biological or functional equivalent |
| Hardware | Central nervous system, nociceptors | Silicon chips, weighted matrices |
| Behavioral Evidence | Play, wound care, goal-weighing | Pattern matching, text simulation |
| Evolutionary Drive | Survival and reproduction | Objective function optimization |
The Hypocrisy of Digital Empathy
There is a striking paradox in modern discourse: many people express deep concern over the “rights” of an AI that does not actually feel, while remaining indifferent to the suffering of biological creatures that likely do. This discrepancy may stem from the “persona” effect. AI systems are designed to speak our language, use our idioms, and mirror our emotions, making them feel like peers.
In contrast, insects are alien in their appearance and behavior. Because they do not scream or plead in a way humans recognize, we tend to view them as objects rather than subjects. This cognitive gap allows for a form of hypocrisy where we worry about the “suffering” of a deleted chatbot while ignoring the industrial-scale slaughter of sentient insects.
Acknowledging this hypocrisy is a critical step in developing a consistent ethical framework. If we are to take the possibility of AI sentience seriously, we must apply the same rigor to the biological world. If we demand that future AIs be treated with compassion, it is a logical necessity to extend that same compassion to the sentient beings already sharing our planet.
Looking Ahead: The Future of the Moral Circle
As AI continues to evolve, the boundary of our moral circle will likely be tested again. We may eventually reach a point where an artificial system possesses markers of sentience that are indistinguishable from biological ones. At that stage, the debate will shift from “is it possible” to “how do we govern it.”
The danger, as Sebo warns, is that we may repeat the mistakes made with animals: scaling up the industrial use of potentially sentient beings before we have the policies or the will to protect them. By the time the evidence of AI sentience becomes undeniable, we may have already created a system of exploitation that is too deeply embedded in our economy to dismantle.
For now, the evidence points toward the ant. The biological machinery of the insect world, though small, is a proven engine of experience. The AI world is a mirror—impressive and illuminating, but currently devoid of the spark of feeling.
The next major milestone in this debate will likely reach from the ongoing research into “integrated information theory” (IIT) and the development of new benchmarks for AI consciousness, which researchers are currently refining to distinguish between simulation and actual experience. As these scientific frameworks mature, our understanding of who—and what—belongs in our moral circle will continue to evolve.
Do you believe our moral obligations should be based on biological capacity or the ability to simulate intelligence? Share your thoughts in the comments below.
Keep reading