Joe Abercrombie once wrote, “If you want to change the world, you must first understand it.” In his novels, this line is aimed at magic, but he continues: “A carpenter first seeks to understand wood, a smith must understand metal; one cannot change the world before they understand the principles of what they are working with.”
It is a sentiment that could be applied to almost any field, but nowhere is it more relevant than in the life sciences.
In the pharmaceutical industry, the act of understanding is phenomenally hard. Biology is fractal in its complexity. At the human level, we see outcomes: disease, symptoms, biomarkers. Drop to the cellular level and we enter a tangle of thousands of interacting proteins, lipids, and RNAs in constant motion. Simplify again to a single protein and new layers of complexity emerge as flexible structures breathe and binding pockets shift.
Then physics enters the scene. Free energy perturbation calculations, molecular dynamics simulations and docking studies all attempt to capture the dance of atoms in motion. Each level of abstraction simplifies, but also erodes context. A molecule that binds perfectly in vitro may cause toxicity in animals.
The deeper we look, the harder the truth becomes to hold onto.
I remember the first lecture of my biochemistry and genetics degree. The professor opened with a smile and said,
“Everything you have been taught so far was a gross oversimplification. Welcome to the truth. It only gets worse from here.”
They were right. The more we learn, the more we realise that certainty is a luxury of the uninformed. Progress in science is not about conquering complexity but about learning to navigate it with humility and precision. You need to know your arsenic from your elbowium.
The same lesson applies in machine learning. Models appear to “understand” chemistry or biology, but most of the time they are only finding shortcuts. They learn to cheat, exploiting correlations in the data that minimise error rather than representing any underlying mechanism.
A model might predict a compound’s activity because it recognises a molecular weight pattern, not because it has captured thermodynamics or binding energetics. Without deep domain insight, such models are statistical tricksters that confuse correlation with causation.
At Ignota Labs we face this daily. Our work sits at the intersection of mechanistic toxicology and machine learning, where both rigour and scepticism are essential.
That is why our scientific team is built from people who have lived the complexity, not just read about it. Every member of our science team holds a PhD and has years of experience in drug discovery. Our technology team includes Cambridge-trained computational scientists and AI specialists, supported by experienced industry critics whose role is to ensure our models reflect biological reality rather than mathematical convenience.
Trying to change a system you do not understand is not innovation; it is arrogance. G. K. Chesterton captured this in his famous parable of the fence: when you find a fence in the road and do not see its purpose, the worst thing you can do is tear it down. You must first understand why it was built before deciding whether it should remain.
In biotechnology, would-be revolutionaries attempted to tear down fences without understanding why they were there, and hit immovable regulatory walls. Safety regulations, clinical trial design, medicinal chemistry discipline, and data scepticism each exist for a reason. They may be imperfect, but they are the scaffolding that keeps our world from collapsing under its own complexity.
AI Drug Discovery has just come out of a huge hype cycle which burst in the last 2-3 years. As the next generation of companies emerge, we can learn from previous mistakes, and find the plateau of productivity that will actually make a lasting impact on the industry.
At Ignota Labs, we innovate from a position of deep understanding. Our turnaround model is rooted in identifying where progress was previously blocked, understanding why those limits were set, and determining whether they still hold true. Some boundaries were essential safeguards, others were built on outdated assumptions. By learning the difference, we can move or remove the barriers that no longer serve science, changing the paradigm through informed innovation rather than blind disruption.
Whether you are building a molecule, a model or a company, the principle holds true: you cannot change the world until you understand it. Biology, AI and business each have their own laws, limitations and rhythms. To innovate, you must first learn.
At Ignota Labs, that is our philosophy. We seek to understand before we act. Only then can technology serve science, and science serve humanity.
Author: Dr Jordan Lane
