How AI Unlocks the Secrets of Life (AIBIO-UK Mini-Series)

Charlie Harrison
Jul 1
5 min read

Updated: Sep 17

Welcome to the cutting edge of science! For 20 episodes, we’ve explored the fascinating world of AI, but today, we’re diving into an area that’s not just important but rapidly developing: the intersection of Artificial Intelligence and Biology. This isn’t just a niche topic; it’s a field where incredible research is emerging, potentially bringing once-futuristic concepts into reality. Imagine curing genetic diseases in children, turning leftover food into car fuel, or creating plastic bottles that biodegrade on their own. All these possibilities stem from the integration of AI and biology, offering a path to reprogram genetic code (DNA) as easily and cheaply as editing computer software.

by Charlie Harrison

Listen Now

Why AI and Biology Are a Perfect Match

The sheer scale and structure of biological data are too vast for human comprehension. A single human genome contains three billion bases – the building blocks of DNA – and a human has 30 trillion cells.

If you unraveled all the DNA in your body and laid it end to end, it would stretch from the sun to Pluto and back 17 times.

This truly is a “big data set”, and AI is uniquely capable of navigating this immense scale of information.

Revolutionary Applications of AI in Biology

The impact of AI in biology is already profound, transforming various sectors:

Healthcare and Drug Design: AI excels at digesting and analysing enormous amounts of health data to find patterns. During the COVID-19 pandemic, Pfizer used AI and machine learning to reduce clinical trial data analysis time from months to just 22 hours, significantly speeding up vaccine development. AI is now being applied much earlier in the drug development pipeline to generate and screen potential new drugs and drug combinations before laboratory testing or clinical trials, saving immense amounts time and resources. A company called Exscientia (now Recursion), which prioritises AI in drug development, managed to reduce their exploratory research phase for an immuno-oncology drug from four and a half years to just 12 months. Digital twins promise to personalise medicine, producing tailored treatments optimised for an individual rather than a population.
Tackling Superbugs: AI is not just speeding up existing processes; it’s enabling entirely new discoveries. In May 2023, scientists used AI to discover a brand new class of antibiotics capable of killing deadly antibiotic-resistant bacteria. These antibiotics were generated “from scratch” by AI, in ways that conventional biologists might not have considered.
Environmental Sector and Sustainability: AI has been used to generate novel plastic-eating enzymes by predicting optimal adaptations and mutations to existing natural enzymes. This process is akin to AI “speeding up evolution,” finding the most effective adaptations for consuming plastic. AI is helping to predict how to design and genetically engineer living organisms (like bacteria or enzymes) to sustainably produce biomaterials, biofuels, and biopolymers. This includes concepts like using leftovers to fuel cars, creating sustainable plastics, or capturing carbon dioxide from the atmosphere to converting greenhouse gas emissions into valuable raw materials.

The Journey: How AI Met Biology

The integration of AI and biology wasn’t always obvious. Historically, these were separate academic fields. The timeline of their convergence can be split into four key stages:

Before Computers: Early applications involved using mathematical equations to model biological processes, such as population dynamics (e.g., rabbit and fox populations) and the formation of spotted and striped patterns through diffusion, demonstrating that some biological phenomena could be explained mathematically.
The Arrival of Computers: This era was marked by the creation of large biological datasets. The Genbank database, a “Google for genetic information,” allowed for the digital storage of vast genetic sequences that wouldn’t fit on paper. Simultaneously, the development of high-throughput robotics and measurement techniques in laboratories automated previously painstaking experiments, leading to an explosion of data that could be stored in databases like Genbank.
The Machine Learning Era: With massive datasets available, research groups began applying machine learning, which naturally excels at finding patterns in large data. The most revolutionary application came with Google DeepMind’s AlphaFold. Proteins are life’s fundamental building blocks, and their 3D shape dictates their function. AlphaFold can take a protein sequence and predict its complex 3D structure, a task that would previously have absorbed an entire PhD project. AlphaFold has now published over 200 million protein structures, saving billions of research hours.
The Generative AI Era (Today): Building on predictive models like AlphaFold, generative AI models like Evo can now create entirely new biological entities. For example, they can generate new protein designs, which AlphaFold can then evaluate for feasibility and properties. This combination of generation and prediction will massively accelerate the exploration of different biological structures, leading to the creation of new drugs, plastic-eating enzymes, and lab-grown foods.

Ethical Considerations: High Stakes and Future Challenges

While the potential benefits are immense, the intersection of AI and biology comes with significant ethical considerations. The stakes are incredibly high, especially when dealing with global problems like pandemics and diseases.

Dual Use Technology: Any technology with such transformative potential also carries a risk of harm. AI in biology is a “dual use” technology: tools that can save lives could also be used to create bioweapons or dangerous pathogens with no known cures. This concern about AI for bioweapons is often overlooked in broader discussions about AI’s existential threats.
AI Interpretability: A challenging aspect is that we often don’t know exactly what complex AI models like AlphaFold have learned. There’s a trade-off: giving AI the flexibility to understand patterns beyond human comprehension often makes the outputs less understandable to us. While methods like “mechanistic interpretability” are being developed, it remains a difficult challenge.
Public Discourse and Privacy: There’s an urgent need for more public, governmental, and scientific discourse on these applications. A major concern is the use of human medical data and genomes to train AI, raising significant privacy issues.

The Future: Utopia or Inequality?

Looking ahead, AI in biology envisions some truly transformative futures:

Precision Medicine: This concept aims for an AI model of each individual’s cells and genome, allowing for customised treatments tailored to a person’s specific biology. This could lead to higher efficacy and fewer side-effects compared to current general treatments that are developed over populations.
Harnessing Biology for Creation: The ultimate goal is to understand and manipulate biology at every level – from reactions to protein folding – to produce new things and grow cures. This could mean growing your own medicine at home or even producing food. The hope is that this capability could make things more abundant and accessible, democratising production that was previously expensive and resource-intensive.

However, a critical question remains: will these amazing benefits exacerbate inequality? There’s concern that high costs could make technologies precision medicine only accessible to the wealthy, creating a genetic lottery where the rich have an advantage. While some applications, like sustainable fuel from leftovers, might naturally become cheaper and more accessible, the commercialisation and gatekeeping of medical innovations, particularly within privatised healthcare systems, present a significant challenge that will require careful consideration and systemic changes.

This first episode of the AIBIO-UK mini-series offers a comprehensive overview of how far we’ve come – from “pencil and paper” biological models to the generative AI that is reshaping life itself. Stay tuned for future episodes that will dive deeper into specific topics, including how AI decodes genes, the full story of AlphaFold, the future of AI pharmacists, AI’s role in saving our planet, and the crucial ethical path forward.

If you enjoyed reading, don’t forget to subscribe to our newsletter for more, share it with a friend or family member, and let us know your thoughts—whether it’s feedback, future topics, or guest ideas, we’d love to hear from you!