AI has captured our collective imagination, promising to revolutionize scientific research, healthcare, education and medicine. The headlines are compelling: AI designs new drugs in months instead of years; algorithms decode the mysteries of neural networks; machine learning accelerates the path from laboratory to patient.

This enthusiasm follows a familiar pattern. Every new technology begins with euphoria before settling into realistic evaluation. Elon Musk’s recent prediction that AI will surpass human intelligence by 2026 and all human intelligence by 2030 exemplifies this tendency. Such pronouncements, while attention-grabbing, often reflect Silicon Valley optimism rather than scientific consensus about the pace of AI advancement.

Where AI is actually delivering

During the COVID-19 pandemic, AI helped identify promising drug candidates and accelerated vaccine development timelines. Large language models are now scanning millions of research papers to identify potential therapeutic connections that would take human researchers years to discover.

Meanwhile, in neuroscience, AI is being used to decode brain signals from paralyzed patients, enabling them to control computer cursors and robotic arms with unprecedented precision. Brain-computer interfaces powered by machine learning are translating neural activity into text, giving voice to patients who have lost the ability to speak. Researchers are using AI to map neural circuits with cellular precision and simulate brain networks that were previously too complex to model.

In structural biology, AI has achieved remarkable breakthroughs in protein structure prediction, which have major implications in drug discovery. Google DeepMind’s AlphaFold can now predict how proteins fold with stunning accuracy, solving a puzzle that has stumped scientists for decades. This matters because understanding protein structure is fundamental to developing new treatments for human diseases.

In drug discovery, we’re seeing real progress too. Companies like Exscientia made history with the molecule DSP-1181, the first AI-designed drug to enter human clinical trials for treating obsessive-compulsive disorder. In-silico Medicine became the first company to advance an AI-designed drug for an AI-discovered target into clinical trials — a “double first” where AI handled both target identification and drug design. Others, like Recursion Pharmaceuticals, have used AI to identify new drug targets and advance candidates like REC-1245 (an orally bioavailable molecular degrader of the RNA-binding protein 39) for solid tumors from discovery to pre-clinical testing in just 18 months, less than half the typical timeline.

But here’s what breathless media coverage misses: these are incremental improvements in specific, well-defined problems, not the wholesale transformation of medicine that venture capitalists and tech evangelists would have you believe.

Hype and its consequences

The AI drug discovery sector has attracted billions in investment, with startups promising to turn drug development from an uncertain, lengthy process into something resembling software engineering that is predictable, systematic and fast.

Companies like Insitro have raised $700 million in venture funding, while hundreds of AI drug-discovery startups have collectively raised billions more. This hype creates what one pharmaceutical researcher calls “FOMO” (fear of missing out) among decision makers who worry they’ll be left behind if they don’t embrace AI.

The problem is that overhyping AI creates unrealistic expectations. When a pharmaceutical executive hears that AI will “solve all our problems,” they expect magic. The reality is more sobering: while about 20 AI-discovered drugs are currently in clinical trials, none have yet received Food and Drug Administration (FDA) approval. When the reality proves more modest, with AI helping with specific tasks rather than revolutionizing entire processes, disappointment follows. As one computational chemist put it, “Every time someone says deep learning, I bring out my magic wand.”

This disappointment isn’t just about hurt feelings. It can set back the entire field. The tech industry has seen this pattern before — the “AI winters” of the 1970s and 1980s occurred when artificial intelligence failed to meet inflated promises, leading to dramatic funding cuts that lasted for decades. Researchers worry that when AI fails to meet inflated promises, funding will dry up and legitimate applications will be abandoned along with the hype.

The scientific reality check

AI currently excels when it has large, high-quality datasets with clear patterns to recognize. Protein structures fit this bill perfectly, as there are only 20 amino acids, and proteins follow predictable rules. That’s why AlphaFold works so well.

Drug discovery for small molecules is far messier. The chemistry is vast, the data is often inconsistent or wrong, and human factors play a huge role. As one industry veteran noted, the synthetic organic chemistry literature is a mess, full of biases and inaccuracies that AI models inadvertently learn.

The biggest challenges in drug development are selecting the right targets and predicting human toxicity; however, these challenges remain largely beyond AI’s current reach. These problems kill most drug programs. They also require the kind of biological insight and creative reasoning that AI hasn’t mastered.

In neuroscience, AI offers exciting possibilities for modeling brain activity and understanding neural networks. Researchers are training machine learning systems on vast datasets of brain connectivity maps and neural recordings, hoping to simulate brain function in ways that traditional mathematical models cannot.

But the brain presents unique challenges. Unlike proteins, neural networks are incredibly variable between individuals and constantly changing. The data required to train AI models must come from the same specimen because you can’t mix brain connectivity maps from one animal with neural activity recordings from another. This creates logistical nightmares and limits the scope of what’s currently possible.

Looking ahead     

The next five years will likely separate the winners from the losers. We’ll see some AI drug discovery companies deliver genuine successes, while others will quietly change direction or shut down. Companies like Exscientia and Recursion, which focus on specific problems like drug design or target identification, are more likely to succeed than startups promising to solve all of drug discovery at once.

For example, we should expect to see the first AI-designed drugs receive FDA approval within the next few years, given that about 20 such drugs are currently in various stages of clinical trials. However, the failure rate will remain high. Most AI drug discovery companies will likely discover that their algorithms work better for some diseases than others, forcing them to narrow their focus.

In brain research, we should expect steady progress in specific areas, such as brain-computer interfaces for paralyzed patients and better computer models of simple brain circuits. AI is also showing promise for developmental conditions like cerebral palsy and autism. For cerebral palsy, researchers are developing AI-powered gait analysis systems that can optimize physical therapy and predict which treatments will work best for individual patients. Brain-computer interfaces may eventually help people with severe motor disabilities control assistive devices more naturally.

For autism, AI tools are being developed to detect early signs through video analysis of infant behavior, potentially enabling earlier intervention. AI-powered apps are also being tested to help with social skills training and communication, though these remain largely experimental. However, breakthrough treatments for complex brain diseases like Alzheimer’s or schizophrenia will take much longer. The human brain is not a computer, and treating it like one has limitations.

Most importantly, we’ll likely see a reality check in expectations. Companies will stop claiming they can revolutionize all of medicine and start focusing on solving specific problems well. Investors will become more careful about which AI health companies they fund. The current wave of “AI will solve everything” marketing will give way to more honest discussions about what AI can and cannot do.

The path forward

AI is not a silver bullet for medicine’s greatest challenges. The key is applying it strategically to problems where it has genuine advantages while maintaining realistic expectations about what it can accomplish.

The most exciting developments will likely come from hybrid approaches that combine AI with traditional scientific methods, leveraging the strengths of both. Think of AI as a powerful new instrument in the researcher’s toolkit, not a replacement for human insight and creativity.

The real promise of AI in medicine lies not in its ability to replace human scientists, but in its capacity to augment human intelligence, helping us see patterns we might miss and explore possibilities we hadn’t considered. That’s a more modest vision than the grand promises of Silicon Valley, but it’s also a more achievable one — and ultimately more valuable for patients waiting for new treatments.

[Dr. Mohammad Farhan is an Associate Professor at the College of Health and Life Sciences at Hamad Bin Khalifa University.]

[Hamad Bin Khalifa University’s Communications Directorate has submitted this piece on behalf of its author. The thoughts and views expressed are the author’s own and do not necessarily reflect an official University stance.]

[Kaitlyn Diana edited this piece.]

The views expressed in this article are the author’s own and do not necessarily reflect Fair Observer’s editorial policy.