Can artificial intelligence prevent the next pandemic, help crack the puzzle of Alzheimer’s disease, or speed up the development of treatments for cancer? The answer is no longer theoretical.

Today, around 170 drugs discovered or designed with the help of AI are in clinical trials. As a result, 2026 and 2027 are shaping up to be milestone years, as the real-world capabilities and limitations of this technology in one of science’s most critical fields will be put to the test.

Early signs are encouraging. Just days ago, researchers at the University of Cambridge announced the first results from a new vaccine technology that aims to offer protection not only against known viruses, but also against future variants and related viruses that have not yet appeared in humans.

The “Super-Antigen”

The innovation relies on AI and machine learning to design what researchers are calling a “super-antigen.” To do so, scientists analyzed thousands of coronavirus genetic sequences and identified common features that remain stable despite mutations. Based on these, they created a synthetic antigen capable of priming the immune system against entire families of viruses.

Initial trials involving 39 healthy volunteers showed the vaccine to be safe and well tolerated. The technology is still at an early stage, however, and a larger Phase 2 study with around 200 participants is planned to follow.

“This development marks a significant philosophical shift: from developing vaccines as a response to a pandemic, we are moving toward the proactive design of vaccines that could offer protection against future threats,” Maria Gazoulei, professor of Biology, Genetics, and Nanomedicine at the University of Athens Medical School, told the paper. “While AI cannot replace the necessary stages of laboratory and clinical evaluation, it can significantly reduce development time and strengthen global readiness against future pandemics.”

Ten Years Saved

Meanwhile, a growing number of researchers worldwide are finding AI to be an invaluable collaborator. Scientists are using it to scan existing drugs, including treatments for diabetes and heart disease, to determine whether they might also block the accumulation of brain-damaging proteins, specifically beta-amyloid and tau, which are responsible for triggering Alzheimer’s disease. Perhaps most strikingly, this approach has allowed researchers to save what would have amounted to roughly ten years of analysis and trials.

As Gazoulei clarifies, AI has long since outgrown its role as a simple data analysis tool. “Today it can identify patterns that escape human attention, propose new hypotheses, design new molecules, materials or proteins, and accelerate scientific discovery across many fields.”

Beyond coronaviruses, the Cambridge team is also using AI to identify the “stable” regions of the influenza virus, including H5N1 bird flu, in hopes of developing a vaccine that would not need annual reformulation. Similar approaches are being applied in the search for an effective Ebola vaccine.

Innovative Treatments

Researchers are also pushing forward on novel therapies. “Although no drug developed exclusively by AI has yet received full regulatory approval, several AI-designed drug candidates are already in clinical trials,” Gazoulei notes.

One of the most cited examples is Rentosertib (INS018_055), the first drug whose therapeutic target and molecule were both identified using generative AI. Developed for idiopathic pulmonary fibrosis, it successfully completed a Phase 2a clinical study with a satisfactory safety profile and encouraging early efficacy data. “This development shows that AI is no longer merely a research tool, but a genuine accelerator of pharmaceutical innovation.”

Just days ago, Japanese pharmaceutical company Takeda also announced that a psoriasis drug developed with the help of AI outperformed an already approved treatment in a clinical trial, making it one of the first cases where AI appears to have led not just to a new drug, but to a better one.

Trials are expanding further into personalized medicine, with new multimodal AI models being evaluated for their ability to predict which drug combinations are most likely to work for specific patients and which are likely to cause serious side effects.

Myths and Facts About AI

Could AI ever replace human beings? That is perhaps the biggest myth, according to Gazoulei. “In reality, it can perform certain tasks very quickly and efficiently, but it lacks the experience, intuition, empathy, and judgment that characterize human thought.”

Another common misconception is that AI is always right. While it can process enormous volumes of information in a fraction of the time, it still makes mistakes, reproduces biases embedded in its training data, and can produce answers that sound correct without actually being so.

“AI can indeed significantly increase productivity, particularly in tasks requiring speed and data analysis. But productivity is not measured only by how fast something is done, but also by how correctly, creatively, and responsibly it is done. The limitations we see today will likely not be the same in ten or twenty years,” she adds. Whether machines will ever achieve genuine understanding, consciousness, or moral judgment as humans possess remains an open question. “For now, at least, we are not talking about a battle of human versus machine, but about a new form of collaboration. Humans bring creativity, judgment, and values; AI brings speed and computing power. The combination of the two is likely the most powerful tool we have at our disposal.”