In common misuse scenarios, these technologies could lead to the development of targeted bioweapons that pose significant threats to global security. This practice could lead to new forms of discrimination and inequality, as those who can afford genetic enhancements may gain unfair advantages. Addressing these challenges requires a thoughtful approach, balancing the pursuit of human enhancement with a commitment to ethical principles, equity, and the preservation of our shared humanity. This disparity could deepen social divides and create a new class of genetically privileged individuals. These enhancements would not only boost individual potential but also contribute to societal advancements by accelerating innovation and efficiency. But training on the tool to design an experiment is complicated and time-consuming — even for seasoned scientists. The technology, CRISPR-GPT, acts as a gene-editing “copilot” supported by AI to help researchers — even those unfamiliar with gene editing — generate designs, analyze data and troubleshoot design flaws. These may well be wrong, though the degree to which they are wrong will only be revealed with time – these are provided to be intentionally provocative and hopefully lead to useful discussions that may ready our field for change. But critics fear the research opens the way for unscrupulous researchers seeking to create enhanced or modified humans. Over 80% of genome-wide association studies to date have been conducted on individuals of European ancestry (Schumacher-Schuh et al., 2022), leading to predictive tools that underperform in other populations. To comprehensively analyze the usage patterns of transformer-based models in genetics and hereditary diseases, a systematic review approach was developed according to the latest PRISMA 2020 guidelines for reporting systematic reviews (Page et al., 2021), ensuring thorough and transparent coverage of relevant studies. Recent reviews highlight the growing impact of ML approaches in biomedical fields, including applications in diagnosing rare diseases and improving clinical outcomes (Manjurul Ahsan et al., 2022; Roman-Naranjo et al., 2023). "These advancements are uncovering biomarkers and therapeutic targets missed by traditional protein-coding approaches," Pupo noted. Generative Adversarial Networks (GANs) (Goodfellow et al., 2014) complement this landscape as specialized models for data generation, enabling the synthesis of highly realistic images, biomedical data, and even artificial genetic sequences through adversarial training. Vision Transformers (ViTs) have further extended this approach beyond text, applying the transformer architecture to image processing tasks (Dosovitskiy et al., 2021). This adaptability enables LLMs to be highly efficient across a range of applications, including research, healthcare, and education, without the need for retraining from scratch (Du et al., 2024; Zampatti et al., 2024; Aronson et al., 2024; Ueda et al., 2024; Laye and Wells, 2024). These models are capable of performing a variety of tasks by leveraging either full training on large datasets or fine-tuning with smaller, task-specific datasets. But the demand for other jobs like machine learning specialists and data center technicians has risen. Businesses may transition from a model of human-led workflows to a hybrid workforce where humans act as orchestrators for AI agents. In particular, advancements in AI agents will enable people to hand off more complex tasks to automation. Company leaders don’t have to spend time parsing through the data themselves, instead using instant insights to make informed decisions. AI’s ability to analyze massive amounts of data and convert its findings into convenient visual formats can also accelerate the decision-making process. AI has also been used to help sequence RNA for vaccines and model human speech, technologies that rely on model- and algorithm-based machine learning and increasingly focus on perception, reasoning and generalization. The research case in favour of pursuing germline gene editing is very strong.1 Editing human embryonic stem cells could be a breakthrough for the study of early human development. It is crucial that we make sensible decisions about the development and use of gene-editing technologies. The ability to alter our biological makeup will create immense opportunities but also pose novel threats. In this unsupervised ML approach, we would not label the individuals as having specific conditions such as forms of MODY; there would be no “ground truth”. However, these rules-based approaches may still be combined in very useful ways with the more modern AI methods described below. This misuse could lead to scenarios where individuals’ genetic data is exploited for discriminatory purposes or financial gain. As an upshot, while medicine will change dramatically due to AI, exactly what the future of medicine holds – and how these changes will affect patients, clinicians, researchers, and society – remains murkier. This loss of personal freedom and autonomy could lead to a society where genetic determinism dictates one’s opportunities and quality of life, eroding the fundamental principles of individual rights and freedoms. In the most dystopian extreme, individuals could be forced to undergo genetic modifications or be constantly monitored and controlled based on their genetic profiles. To enhance genomic data aggregation for analysis, cloud providers are playing a key role, Pupo noted. Large language models could potentially translate nucleic acid sequences to language, thereby unlocking new opportunities to analyze DNA, RNA and downstream amino acid sequences, said Aber Whitcomb, CEO of AI development and platform provider Salt AI. There have been considerable advancements in the algorithms needed to make sense of genomics data. "It is clear that ongoing advancements in high-throughput instruments have decreased the cost of sequencing, which in turn increases the amount of data available for analysis," said Jeff Elton, CEO at SaaS and data platform provider ConcertAI. Eight latent topics were extracted using Latent Dirichlet Allocation (LDA), capturing distinct research themes from clinical variant interpretation to computational method development. The fine-tuned analysis showed similar patterns with overlap (), confirming the robustness of these findings. The fine-tuned analysis confirmed that domain-specific trends remain stable across different filtering strategies, validating our findings. Despite this, the Trump administration’s AI Action Plan unveiled in 2025 emphasizes a largely hands-off approach to AI regulation. For example, there have been incidents where AI systems reinforce gender roles or AI image generators overwhelmingly create images of white men when prompted to show an image of a successful person. Some experts are concerned that by using personal data, AI systems are able to infer sensitive information such as sexual orientation, political views or health status, and that this could lead to algorithmic bias or the stereotyping of certain groups. Even though some AI companies have agreements with certain platforms to train on their data, individuals whose information might be included in those agreements have no say on how their personal data is used. In genomics and precision medicine, the lack of diversity in training data has long limited the generalizability of AI insights for underrepresented groups. Despite their impressive capabilities, LLMs often reflect biases present in their training data, which can affect clinical utility. Equally important are the integrity of training and evaluation datasets, the representativeness of benchmarks, and the methods used to integrate and align multimodal inputs. Together, these developments underscore that the value of LLMs in genetics is not solely defined by model architecture. Recent work in other technical domains has highlighted the threat of benchmark leakage, where models inadvertently see test data during pretraining (Zhou et al., 2025; Ni et al., 2025). Predicting exactly how medicine, including clinical genetics, will evolve due to AI is difficult due to many regulatory, legal, financial, logistic, ethical, computational, and other factors, and the fact that many different stakeholders disagree about optimal approaches and outcomes. The introduction of genetically enhanced individuals could lead to significant inequalities, as those with access to such technologies would gain unprecedented advantages over the naturally born population. Such enhancements could lead to a new era of exploration and achievement, enabling humans to thrive in previously inhospitable environments, such as deep ocean depths or distant planets. 🚫 Will AI-driven gene editing create a genetic divide between enhanced and non-enhanced humans? Although newborn screening of bloodspots taken from the heel to identify a few dozen rare diseases has been going on for decades, today’s AI-driven approaches are taking diagnosis of the ultra-rare to a new level. https://www.dnaxplore.com/ occur when these genes go wrong so the studies could lead to better treatments, according to Prof Matthew Hurles, director of the Wellcome Sanger Insititute which sequenced the largest proportion of the Human Genome. This collaboration is pivotal in advancing the integration of tabular-to-image converter models with CNNs, thus propelling the horizons of omics data analysis and interpretation. The optimal gene locations are then adjusted for coordinate mapping and overlaid onto a grid to create an image representation where each point corresponds to a gene expression value. This composite figure illustrates the transformation of tabular omic data into an image format for analysis using convolutional neural networks (CNNs). This transformative progression in data transformation and analysis signifies a momentous stride forward not only for unraveling the intricate nuances ingrained within tabular data but also for enhancing its predictive modeling capabilities. AI also enables multiomic data analysis beyond genomics for integrated metabolomic, proteomic, transcriptomic and epigenomic analysis. AI has demonstrated the ability to analyze and identify genetic variations and detect patterns and protein structures, she added. "AI is extensively used in genomics to expand its application potential and increase the speed and accuracy at which vast amounts of genomic data are analyzed," said Neeraja V, senior analyst at the consultancy Everest Group. Gradually increasing your training, choosing proper footwear and listening to your body can help keep shin pain from sidelining your workouts “It’s not possible for humans to absorb all of that information at once,” he says. “AI helps reduce those burdens, freeing up time so clinicians can focus more on patients — and patients notice that difference.”