How AI is Changing the Way We Create and Use Medicines

24 Aug 2025

8 Min Read

AP Dr Jason Loo Siau Ee

IN THIS ARTICLE

Contributed by AP Dr Jason Loo Siau Ee, whose research explores computational modelling of proteins and drugs, specialising in structure-based drug design, molecular dynamics simulations, and free energy calculation methods. He can be reached at jasonsiauee.loo@taylors.edu.my.

In the not-so-distant past, discovering a new medicine was like searching for a needle in a haystack: slow, painstaking and often uncertain. Today, that haystack is being searched by a new kind of helper: artificial intelligence. From analysing billions of data points in seconds to imagining molecules no human has ever conceived, AI is transforming how we design the medicines of tomorrow.

To understand where we are headed, it helps to see how drug discovery worked before and how AI is changing the game.

Drug Discovery Then vs Now

The traditional process of drug development is long and resource-intensive, often taking 10 to 15 years. It begins with identifying a biological target, followed by laboratory screening, optimisation, pre-clinical testing in cell and animal models, and multiple phases of human clinical trials. Each stage involves significant trial and error, with many potential compounds failing before approval.

Today, AI is beginning to reshape this early part of the journey. Instead of relying solely on time-consuming lab work to test one possibility after another, algorithms can rapidly analyse massive datasets, model biological interactions, and run virtual screenings of millions of compounds in silico. This allows researchers to predict how molecules might behave, estimate toxicity, and select promising candidates more efficiently. It speeds up the design and early selection phases, though it cannot replace the need for physical lab testing and rigorous clinical trials.

While AI cannot compress the entire development process from more than a decade to just a few months, it can significantly reduce the time needed to identify viable candidates. Stages that once took years—such as hit-to-lead or lead optimisation, can sometimes be completed in months through AI-enabled simulations and predictions.

How AI Designs a Drug

Every AI drug design journey starts with data—lots of it. Biological blueprints such as protein structures and disease pathways, vast chemical libraries, and even past clinical trial results are fed into the system. The richer and cleaner the data, the sharper the AI’s insights.

Once equipped with this information, the AI becomes a hyper-focused detective, scanning hundreds of millions of virtual compounds to predict which ones are most likely to bind effectively to the target and possess the right drug-like qualities. In a matter of hours, it can narrow the list from millions to just a handful of strong contenders ready for lab testing.

_{AI don't just sift through existing molecules—it can imagine entirely new ones. One standout example is REINVENT 4, an opensource AI framework launched in 2024 that harnesses recurrent neural networks and transformer architectures, combined with techniques like reinforcement learning and transfer learning, to design novel small molecules from scratch.}

Notable tools have already reshaped the field. DeepMind’s AlphaFold revolutionised protein structure prediction, earning its creators the 2024 Nobel Prize in Chemistry for AlphaFold2. Deep Docking accelerates structure-based drug design by predicting how potential drugs will fit into a target protein, while DeepTox forecasts possible toxicities long before a compound reaches the lab.

In essence, AI is like a tireless digital chemist—one that never sleeps, never takes coffee breaks, and can run millions of virtual experiments at once, enabling researchers to focus their time and resources on the most promising leads.

Can We Trust Machine-Made Medicine

AI’s ability to design medicines is exciting—but it also raises important questions about trust and safety. One challenge is the so-called ‘black box’ problem: deep learning models can be incredibly accurate, yet their decision-making process is often hard to explain. Scientists may know what the AI predicted but not fully why it reached that conclusion.

Data quality is the foundation for trustworthy AI — and yet, it’s often the weakest link. AI learns from whatever data we feed it, so if the input is incomplete, skewed, or flawed, the output can be misleading. This becomes especially problematic in drug discovery, where datasets may lack diverse examples or underrepresent rare diseases. Common challenges include bias, inconsistent or small datasets, and uneven coverage of conditions — all of which make AI predictions less reliable. The problem is particularly acute in areas like neglected diseases, where data may be sparse or of poorer quality, highlighting the need for human oversight throughout the AI process

_{There’s also the matter of misuse. In 2022, researchers showed that the same AI tools used to discover medicines could be flipped to design toxic molecules, simply by adjusting their goals. This highlights a broader concern in AI, from drug discovery to chatbots like ChatGPT: safety checks are built in by the companies that create these systems, but in theory, determined individuals could find ways to bypass them.}

Finally, there’s the question of regulation. AI-designed drugs still have to go through the same rigorous testing and approval processes as any other medicine: lab experiments, safety trials, and multiple phases of clinical testing. Regulators like the Food and Drug Administration (FDA) or National Pharmaceutical Regulatory Agency (NRPA) don’t approve a drug because it’s AI-made; they approve it because it meets strict safety and effectiveness standards. In other words, while AI may help design the drug, the responsibility for its success or failure still rests with the pharmaceutical company that develops and markets it.

AI can design toxic compounds, raising security and bioterrorism risks. While safeguards can be built in, competition and profit motives may weaken them. Strong, proactive regulation is vital to balance innovation with safety and ensure AI models are not misused.

— Ms Choong Chiau Ling

Who Plays What Role in Tomorrow’s Pharmacy?

In the pharmacy of the future, humans and AI won’t be competing—they’ll be collaborating. Machines excel at tasks that require speed, scale, and precision. They can sift through vast amounts of data in seconds, spot complex patterns invisible to the human eye, and predict how effective a potential drug might be based on chemical structure, patient records, or even genetic information.

But there are things only humans can do well. Even as AI accelerates the discovery of new treatments, pharmacists will remain essential in deciding which therapies truly fit best for each individual. They bring clinical insight, the ability to weigh subtle medical factors, and a deep understanding of how treatments integrate into a person’s life. Beyond this, they are the ones making ethical decisions and providing patient-centred care—listening to concerns, explaining options, and ensuring that medical choices align with someone’s values, lifestyle, and resources.

_{In the near future, pharmacists may work as analytic translators, using AI-generated insights to personalise medicine. For example, an AI system might analyse a patient’s DNA profile to flag potential drug interactions or suggest the safest, most effective treatment options. But it will be the pharmacist who talks to the patient, discusses the risks and benefits, considers whether they can afford the treatment, and ensures they can follow the prescribed plan.}

Rather than learning to code, tomorrow’s pharmacists will need to learn how to use AI effectively—turning its predictions into actionable, compassionate care. The technology will crunch the numbers, but the human touch will make the difference.

Pathways for the Future: What This Means for You

If you’re thinking about a career in pharmacy or life sciences, AI is rewriting the playbook—and opening up brand-new opportunities. In the coming years, we’ll see the rise of interdisciplinary careers like AI Drug Discovery Scientist, Computational Chemist, Bioinformatician. These roles blend science with technology, giving you the chance to work on the cutting edge of medicine.

_{Pharmacy-related degrees are also evolving to match this new reality. Many programmes now weave in data analytics, bioinformatics, and computer-aided drug design alongside traditional pharmacology and chemistry. There’s also a growing emphasis on critical thinking and digital literacy—skills that will help you interpret AI-driven insights and apply them responsibly in patient care.}

One of the most in-demand abilities will be understanding personalised medicine—using genetic profiles, lifestyle data, and medical histories to tailor treatments to individual patients. It’s not just about knowing the science; it’s about connecting that science to real lives, ensuring the right medicine reaches the right person, in the right way, at the right time.

Conclusion

AI isn’t here to replace human. It’s here to amplify what’s possible. In pharmacy and drug discovery, the greatest breakthroughs will come from people and machines working together: AI delivering speed, precision, and powerful predictions, and humans providing insight, empathy, and ethical judgement.

The future of medicine won’t be shaped by algorithms alone, but by the people who know how to harness them—turning complex data into treatments that change lives.

AI is changing how we discover medicines—and pharmaceutical scientists are leading the way. If you’re fascinated by chemistry, biology, and the power of technology, this is where your future begins. Explore how a degree in Pharmacy or Pharmaceutical Science can put you at the forefront of drug innovation.

BOOK AN APPOINTMENT

BACK TO NEWS AND ARTICLES

{{ firstlevel.title }}

{{ firstlevel.title ? firstlevel.title : vm.computedTitle(firstlevel.path) }}

{{ secondlevel.title }}

{{ secondlevel.title }}

{{ secondlevel.title }}

{{ secondlevel.title }}

{{ thirdLevel.title }}