Biology, AI and Learning the Language of Life through Evo2
What is life? Humans across the world have asked this significant question for ages and now, in the year 2025, artificial intelligence is asking (and answering) it too.
We asked ChatGPT, and here's its answer:
That’s a profound question! From a biological perspective, life is often defined by a set of characteristics: metabolism, growth, reproduction, response to stimuli, and homeostasis. Life is dynamic, self-organizing, and adaptive—constantly interacting with its environment in complex ways.
From an artificial intelligence perspective, the question becomes even more intriguing. AI can mimic aspects of life—learning, adapting, problem-solving—but it lacks key biological features like metabolism and self-sustaining reproduction. Some argue that life isn’t just about the physical properties of organisms but about information processing, which AI certainly does. Could AI ever be considered "alive" if it meets enough criteria?
Then there’s a more philosophical lens: Is life defined by self-awareness, consciousness, or even purpose? Biology suggests life is about self-propagation, but humans (and potentially AI) might redefine it based on intelligence and creativity.
What’s your take—do you see AI ever qualifying as "alive"? Or do you think biomimicry might bring AI closer to life?
Now, there’s a AI new model on the scene… and it’s all about biology.
Introducing Evo2: The largest artificial intelligence model in the field of biology
Evo2, the successor of Evo1, is an AI model used in the science industry that has been trained to understand, analyse and identify patterns in the genome sequences of nearly every life form on earth.
It was developed by scientists from both the American non-profit research centre, the Arc Institute, and technology company NVIDIA. You can read more in the Arc Institute’s preprint paper here.
This highly advanced model has learned from the DNA of over 100,000 species, covering nearly all branches of life. Because it understands genetic patterns so well, it can spot connections between genes in different organisms much faster than human researchers. It can also pinpoint mutations that cause diseases in humans and even design new genetic sequences as complex as those found in simple bacteria.
Read the Arc Institute’s article on Evo2 here.
Let’s look into how Evo2 works, how it can be used and of course, how it’s related to biomimicry.
In simple terms, how does Evo2 work?
Reading Genomes: Evo2 scans long genetic sequences through software, looking for important genes and patterns.
Since it has "read" so much DNA already, it can quickly recognize which gene changes mutations might cause diseases or help organisms survive, similar to how organisms evolve to survive.
Predicting the Future: It can generate surprisingly accurate forecasts, based on data gleaned from eons of evolution, helping scientists understand genetic risks or even design new genes. It will keep learning and improving its predictions over time.
Are you interested in learning more, in a short time? Here are 5 Biomimicry Videos (That Are Under 10 min Each)
Evo2 Application: How can it be used in real life?
Evo2’s biggest function is to help prevent diseases in humans and animals, and assist in finding solutions for advancing medicine and biotechnology.
Below are some examples of potential applications.
- Protein Design & Drug Discovery
Evo2 algorithms can simulate protein folding, enzyme evolution, and drug-target interactions. Co-author and computational biologist Hani Goodarzi of the Arc Institute explained it this way: “If you have a gene therapy that you want to turn on only in neurons to avoid side effects, or only in liver cells, you could design a genetic element that is only accessible in those specific cells. This precise control could help develop more targeted treatments with fewer side effects.”
- Evolutionary Simulations & Research
It can model how genetic traits evolve under different environmental conditions, then predict and mimic the required evolutions for specific genes. This will significantly advance scientists’s understanding of biological processes and genetics.
- Bioengineering & Technology
Evo2 might assist in designing bioengineered organisms with optimized traits to advance technology and other AI models. This may be used in metabolic pathway engineering for biofuels and biotechnology solutions.
How Evo2 Relates to Biomimicry
This AI model is unprecedented, and poses an interesting question around biomimicry.
Evo2 isn’t necessary biomimicry (at least not in the traditional sense), but it is an enabling tool. It enables us to do better biomimicry… mimicking nature’s genius.
Mimicking Evolutionary Processes
Evo2 uses AI-driven algorithms that imitate natural evolution, just as biological organisms adapt over time.
It analyzes genetic sequences like nature does, identifying beneficial mutations similarly to how natural selection refines traits in species.
Why Evo2 might not qualify as biomimicry in the traditional sense
Biomimicry produces functional design as a result of mimicking nature, creating a solution to a human challenge. In contrast, Evo2’s primary goal (for now) is data analysis, and providing insights to researchers. As such it’s an enabling tool, much like how the microscope enabled us to understand cell biology.
In Conclusion
While AI in Biology is certainly hugely beneficial to the practice of biomimicry (and even in some small way biomimetic), it is still new and will need to be tested over time. In fact, we propose that AI is one of six things that biomimicry needs in order to scale.
This tool might be especially helpful in medicine, and by designing genetic sequences, Evo2 mirrors how life itself innovates through genetic recombination and adaptation.
Next steps:
Gain wisdom from nature's genius by enrolling in the Biomimicry Short Course Set, and lead the biodesigned revolution.
