Listen again · In Conversation
Can AI Speed Up Drug Discovery? October 16, 2026 · 9:26 · With Maya, Leo, and Alex
We use clear, simple English to explore stories and understand the world. I’m Maya. With me are Leo, our technology and digital life columnist, and Alex, our journalist. Thanks for joining us.
Transcript Tap a sentence to hear it from there.
Maya Welcome to Slow English FM. This is our round table, where we look at one question from more than one side. We use clear, simple English to explore stories and understand the world. I’m Maya. With me are Leo, our technology and digital life columnist, and Alex, our journalist. Thanks for joining us. Maya Our question is whether artificial intelligence, or AI, can help find promising drug candidates faster and at lower cost. A drug candidate is a possible treatment being developed for testing. But finding one is only part of the journey. Leo, what does AI seem to change at the early stage? Leo The report by Channel NewsAsia, or CNA, says Indian scientists and biopharmaceutical companies are using AI to identify possible drug candidates. The reported aim is to do that faster and at lower cost. That could matter because drug development is long and risky. But it is important to keep the claim in proportion: AI may help with part of the work. It does not remove the rest of the process. Maya Alex, can you give us a real example from the report, and explain how far along it got? Alex CNA gives an example from Peptris, an Indian drug-discovery company. Its co-founder, Venkatasubramanian Narayanan, said one of the company’s programmes reached a late preclinical stage in about 18 months. He compared that with more than four years using conventional methods. Maya So it was not yet being tested in people. What does “preclinical” mean here? Alex It means the programme was still before human trials. And the comparison is one company’s account of one programme. It is not proof that AI always makes drug development take 18 months, or that every project will move faster by the same amount. Leo That limit matters, Maya. The example shows why a company might see value in AI: if it can reach a promising candidate sooner, that may reduce some of the time and cost before human testing. But that is an interpretation of the example, not a general result established by it. We should not turn one company’s reported experience into a rule for the whole industry. Maya And even when a candidate reaches human trials, it still has a hard test ahead. Alex, what does the evidence say about that stage? Alex CNA reports that Bugworks, a company developing treatments for drug-resistant infections and cancer, has advanced its lead antibiotic into Phase 1 human trials. Phase 1 is the first stage of testing an experimental drug in people. Its main purpose is to assess safety. Reaching this stage is progress, but it is not the same as winning approval or showing that a treatment works. Maya Leo, why does that distinction matter for the bigger question about AI? Leo Because speed at one stage is not the same as success at the end. A candidate still has to pass through human testing and the wider development process. CNA reports an analysis by the Biotechnology Innovation Organization, Informa Pharma Intelligence and QLS Advisors. It found that 7.9 per cent of drug programmes entering Phase 1 between 2011 and 2020 ultimately received approval. Maya That is a strikingly small share. But it describes programmes from 2011 to 2020, not a prediction about every new drug being developed now. Alex, what does this suggest about the limits of the promise? Alex It shows why reaching a trial is not the finish line. That historical figure does not tell us whether AI will improve the chances for future candidates. It tells us that, in that group from those years, most programmes did not ultimately receive approval. So a faster search for candidates could be useful, but it cannot by itself guarantee a successful treatment. Leo I agree. There are two different questions: can AI help teams find or move a candidate forward more quickly, and can that candidate succeed through the later stages? The first may be possible in some cases, as Peptris’s co-founder says happened with one programme. The second remains uncertain, and the trial data show how difficult drug development can be. Maya There is also a wider push to support this work in India. Alex, what has the government committed? Alex CNA reports that India’s Department of Pharmaceuticals approved about 16 billion rupees, or 168 million US dollars, in September 2026. The funding covers 41 pharmaceutical and medical technology research projects. The report says the projects include new drug discovery and clinical development. That is public support for a range of research; it does not mean all of the money is for AI. Maya Leo, how does that funding fit with the business case for AI? Leo It suggests drug discovery and development are part of a broader research effort. But we should be precise: the funding figure covers 41 projects across pharmaceuticals and medical technology, not just AI drug discovery. The report also gives some industry context. India’s pharmaceutical exports rose 2.1 per cent to about 31 billion US dollars in the financial year that ended in March 2026, according to government data cited by CNA. Maya Does that export figure tell us whether AI is working? Leo No. It gives a sense of the scale of India’s pharmaceutical sector, but it does not show that AI caused export growth, or that the growth came from AI-developed medicines. It is useful background, not evidence of AI’s results. The direct example we have is still the Peptris programme, reported by its co-founder, and it should be understood as a company-specific account. Alex And the Bugworks example gives us a different point in the journey. Its lead antibiotic has reached Phase 1, according to CNA. That is a step into human testing, but the report does not say that this progress was caused by AI. So we should not combine these examples into a claim that AI has already produced a successful medicine. Maya That is a useful distinction. We have one reported comparison about the speed of an early programme, and another example of a drug entering its first human trial. Neither, by itself, answers whether AI can reliably produce approved treatments. Leo, what is the most careful answer to our question? Leo AI may help teams identify promising candidates faster and at lower cost, and the Peptris account gives one example of faster progress to a late preclinical stage. But that is not the whole process. Human trials remain a major test, and the historical approval figure reminds us that many programmes do not reach approval. So AI may speed up part of drug discovery; it does not make the full path quick or certain. Maya Alex, anything you would add? Alex Only that the examples are at different stages. Peptris’s programme was reported at a late preclinical stage. Bugworks’ antibiotic has entered Phase 1. Keeping those stages clear helps us see both the promise and the limits without claiming more than the evidence shows. Maya Today’s discussion leaves us with a balanced answer: Leo explained where AI may help and why one company’s result is not a general rule. Alex set out the examples, the public research funding, and the challenge of human trials. AI may help speed up an early part of drug development, but approval remains a difficult and separate test. Thanks for listening to Slow English FM.
