Artificial intelligence is making waves across industries, but its impact is higher in some sectors than others. Medicine and other sciences stand to gain much from this technology, thanks to their data-heavy work and demand for speed and accuracy. In these fields, gene editing is a particularly promising use case for AI.
The practice of modifying genes to control specific outcomes in living organisms first appeared in fiction, but it emerged in real-world experiments around the 1960s. Over the decades, it has evolved to produce several cutting-edge medical breakthroughs and research possibilities. Still, scientists have only scratched the surface of what gene editing can accomplish. AI could be the next big step.
How AI Is Changing Gene Editing
Researchers have already begun experimenting with AI in gene research and editing. Despite being a relatively new concept, it’s already yielded impressive results.
Increased Gene Editing Accuracy
One of AI’s most notable advantages in gene editing is its ability to improve this process’ accuracy. Classifying which genes produce which changes is crucial to reliable gene editing but has historically been complex and error-prone. AI can identify these relationships with additional precision.
A 2023 study developed a machine learning model that achieved up to 90% accuracy in determining whether mutations were harmful or benign. This insight helps medical professionals understand what to look for or identify which genes to treat to prevent given health outcomes.
Accuracy in gene editing is also a matter of understanding complex relationships between DNA and proteins. Using the appropriate protein structure is essential when attaching to and removing gene sequences. Scientists have recently found that AI can analyze 49 billion protein-DNA interactions to develop reliable editing mechanisms for specific genetic strands.
Streamlined Genomic Research
In addition to providing clarity on genomic editing, AI accelerates the process. Predictive analytics models can simulate interactions between various combinations of genetic material far quicker than real-world manual testing. As a result, they can highlight promising research areas, leading to breakthroughs in less time.
This AI use case helped biopharma companies deliver COVID-19 vaccines in record time. Moderna produced and tested over 1,000 RNA strands per month when manual methods would’ve only created 30. Without the speed of machine learning, it likely would have taken much longer to recognize which genetic interactions were the most promising for fighting COVID-19.
These applications can drive results outside medicine, too. Predictive analytics can model gene editing possibilities to suggest ways to modify crops to make them more climate-resilient or require fewer resources. Accelerating research in such areas would help scientists make needed improvements to mitigate climate change before the worst effects take hold.
Personalized Medicine
Some of the most groundbreaking uses of AI in gene editing take it to a more focused level. Instead of looking at broad genetic trends, machine learning models can analyze specific people’s genomes. This granular analysis enables personalized medicine — tailoring genetic treatments to the individual for better patient outcomes.
Doctors have already begun using AI to analyze protein changes in cancer cells to pinpoint which treatment would be the most helpful for a specific case. Similarly, predictive analytics can account for patients’ unique genetic makeup, which can influence treatment efficacy, side effects or the likelihood of some developments.
When health care systems can tailor care to the individual on a genetic level, they can minimize unwanted side effects and ensure they pursue the best treatment first. As a result, more people can get the help they need with fewer risks.
Potential Issues With AI in Gene Editing
As promising as these early use cases are, the application of AI in gene editing carries some potential pitfalls. Viewing these dangers in light of the benefits can help scientists determine how best to apply this technology.
High Costs
Like many new technologies, the advanced AI systems necessary for gene editing are expensive. Gene editing is already a cost-prohibitive process — some gene therapies cost as much as $3.5 million per treatment — and machine learning may make it more so. Adding another technology cost might make it inaccessible.
This financial barrier raises ethical questions. Gene editing is a powerful technology, so if it’s only available to the wealthy, it could widen the existing gap in care equality. Such a divide would harm the health of working and middle-class families and become a social justice issue.
On the other hand, AI has the potential to reduce costs, too. Streamlined research and fewer errors could lead to faster technological development and justify lower prices on the developers’ end. As a result, gene editing could become more accessible, but only if companies employ AI with this goal in mind.
Safety Concerns
AI’s reliability is another concern. While machine learning is remarkably accurate in many cases, it’s imperfect, but people tend to over-rely on it because of dramatic claims of its precision. In a gene-editing context, this could lead to significant oversights, potentially leading to medical harm or crop damage if people fail to spot AI errors.
In addition to hallucinations, machine learning models tend to exaggerate human biases. This tendency is particularly concerning in health care, where a body of existing research contains historical biases. Because of these omissions, melanoma-detecting AI models are only half as accurate when diagnosing Black patients compared to white populations. Similar trends could have dire consequences when doctors base gene editing decisions on such analysis.
Failing to spot or account for such errors could counteract the primary benefits of personalized medicine, crop augmentation and similar gene editing applications. Reliability issues like these can also be tricky to spot, further complicating the practice.
Where AI Gene Editing Can Go From Here
The future of AI gene editing hinges on how developers and end users can address the obstacles while leaning into the benefits. Explainable AI models will provide a positive step forward. When it’s clear how a machine learning algorithm arrives at a decision, it’s easier to judge it for bias and errors, enabling safer decision-making.
Emphasizing AI for efficiency and error reduction above impressive but expensive processes will help account for cost concerns. Some researchers believe AI could bring gene therapy costs to nearly $0 by removing many of the complications in research, production and delivery. Early experiments have already produced exponential improvements in delivery efficiency, so further advancements could make gene editing accessible.
Ultimately, it depends on what AI gene therapy research focuses on and how quickly the technology can progress. Machine learning could thoroughly disrupt the field if organizations use it correctly.
AI Gene Editing Has Promising Potential
Gene editing has already unlocked new possibilities in medicine, agriculture and beyond. AI could take these benefits further.
While significant roadblocks remain, the future of AI in genetic engineering looks bright. Learning what it can change and what issues it may entail is the first step in ensuring it takes the field to where it needs to be.
