Artificial intelligence (AI) has the potential to revolutionize the process of drug development, making it faster, cheaper, and more accurate. Currently, it takes an average of ten years and $2.5 billion to bring a new drug to market in the United States. However, AI promises to supercharge this process, significantly reducing the time and costs involved in bringing life-saving therapies to market.
As the CEO of Dotmatics, a company that develops software used by millions of scientists globally, I have witnessed firsthand the excitement amongst researchers regarding the possibilities of AI. Having spent nearly two decades in software and technology, I have been collaborating with others towards the point where advancements in technology and science can finally catch up with the vast amount of data generated by scientific research.
So, how exactly can AI transform the effort to create life-saving and quality-of-life-improving therapeutics? By harnessing the power of massive and complex data, researchers can predict how drugs will interact, their toxicity levels, and potential inhibitions. Additionally, AI enables researchers to identify new and successful compounds much more quickly and cost-effectively.
The success of AI in drug development is not merely theoretical. Biotech startups like Relay Therapeutics and Recursion Pharmaceuticals have reported successful clinical trials of drugs developed through AI-powered processes. These drugs have progressed from laboratory and animal studies to being offered to patients in first-in-human trials.
The potential of AI becoming a reality in drug development is thrilling, but it also poses challenges that cannot be ignored. Throughout my career, I have witnessed how leaps in technology transform our lives in expected and unexpected ways. For instance, when I worked in educational technology, we implemented safeguards to prevent plagiarism amongst students sharing study materials. Now, technologies like ChatGPT raise questions about the definition of plagiarism itself.
While there are ongoing debates among politicians and tech executives about imposing a moratorium on AI development, I believe that the answer lies in having honest and actionable discussions about the challenges and necessary safeguards. Every AI expert will have their own opinions on the most pressing concerns. As the CEO of Dotmatics, and reflecting on the transitions I have experienced in different industries, here are some of the questions that come to mind:
Ensuring quality and accuracy are paramount for scientists throughout the drug discovery process and human trials. Given AI’s tendency to hallucinate, which has been well-documented in various large language model systems, how can we ensure that the insights shaping real-world treatments are accurate?
The ethical considerations surrounding AI are vast. How can we leverage the power of genetic data while simultaneously protecting individuals from harm? For instance, what if health insurers had access to certain gene signatures before considering coverage? This raises concerns about potential discrimination or consent-related issues.
At what point will it be appropriate, if ever, to remove humans from the loop? While industries like transportation are moving towards full autonomy, healthcare is approaching this subject with caution. Even companies developing AI for diagnostic purposes without physician input still position their products as aids to physicians, not replacements.
Personally, I find it difficult to envision a future where medical care proceeds without human involvement. If it were to happen, the healthcare industry would need to reimagine everything from patient communication to liability frameworks. However, there is no denying that AI has the potential to handle tasks previously assigned to early-career scientists and technicians.
While researchers are excited to leave behind the burdens of data wrangling, analysis, and annotation, it necessitates changes in education to keep up with these advancements. Curricula should shift towards critical thinking rather than solely focusing on wet lab skills, producing scientists capable of making informed research and business decisions.
Intellectual property (IP) is the lifeblood of the pharmaceutical industry. When AI generates novel drug candidates, the question arises: who owns the IP? Lawmakers are currently grappling with these questions, including the responsibility of AI for patent infringement. The answers to these legal questions will have significant implications for the incentives driving drug discovery.
In the immediate term, the technology community should concentrate on making existing systems AI-ready. While pundits discuss what AI will look like in the next decade, our primary objective should be to ensure that scientists currently using AI have access to a reliable underlying data layer. Cleaning, organizing, and eliminating bias from the data used to train AI is an industry-wide challenge. After all, the promise of AI is only as good as the information it learns from.
We find ourselves on the verge of a radical revolution in how life-saving drugs are developed and utilized. By approaching this crossroad thoughtfully and strategically, working together to address concerns and implement necessary safeguards, the AI revolution, alongside the researchers and scientists using it, will transform scientific discovery for the better.