Interview: Gen AI in life sciences - sorting the hype from the hope

By Liza Laws

- Last updated on GMT

© Getty Images
© Getty Images

Related tags generative AI Artificial intelligence Data management machine learning Research Patient centricity

The Gen AI revolution promises to redefine various industries, with the life sciences sector at the forefront of this transformation. OSP had the chance to speak to Tarun Mathur, CEO at Indegene, who was keen to allay many of the fears those in the industry may have.

It promises to revolutionize the way organizations conduct research, develop life-saving drugs, and engage with customers. However, amid the excitement, it's crucial to discern the genuine progress from the buzz.

What advice would you have for navigating the early days specifically for experimentation vs. restriction?

As Gen AI gains traction, organizations find themselves divided between early adopters and cautious skeptics. While some see Gen AI as a catalyst for innovation and growth, others exercise restraint due to data privacy and security concerns. I had a conversation recently with Matt Anderson, (CDO and Managing Director of The Carlyle Group), he was reassuring that these concerns will be mitigated as the technology rapidly evolves, addressing security issues through advanced data management and generative models. As the technology evolves over what is anticipated to be an extremely short timeframe, nearly all concerns over data security will be addressed via more advanced data management methods and generative models that do not require personal information, or through sandboxed LLMs.

"With Gen AI, you're going to have a real problem coming if your fear is overwhelming your curiosity." - Matt Anderson

For those still pondering the business value, what would you say to them?

Gen AI’s true potential lies in its ability to handle intricate and disorganized datasets and convert them into valuable and actionable insights. It can organize and interpret antiquated data formats into contemporary structures, identify anomalies that might have otherwise gone unnoticed, and illuminate blind spots where crucial information was once absent. Matt emphasized the imperative for organizations to approach their Gen AI endeavors with a spirit of experimentation, and not lose out on the opportunity to harness its transformative power.

Low-hanging fruit and Long-term potential

The world of Gen AI is rife with opportunities, presenting organizations with a diverse array of prospective projects to explore and capitalize upon. Personally, I categorize these opportunities into two distinct areas: the low-hanging fruit and the long-term potential. The low-hanging fruit comprises projects that promise immediate returns on investment, such as optimizing data management practices, enhancing software development efficiency, and addressing prevalent operational challenges. These are the areas where Gen AI can swiftly and tangibly impact businesses, generating rapid results. The long-term and more visionary applications of Gen AI may require more time to fully materialize but possesses the potential to revolutionize entire industries.

Do you think jobs will be created or lost over time?

Among the most promising attributes of Gen AI is its inherent capacity to empower domain experts - individuals who may lack deep technical proficiency but possess deep knowledge within their respective fields. These experts, equipped with the tools and capabilities offered by Gen AI, can now effectively assume the role of software engineers, seamlessly crafting prompts and fine-tuning models to tackle intricate business challenges. This transformative shift in the dynamics of organizational roles not only redefines the traditional paradigms of development but also serves as a catalyst for maximizing human potential, creating new synergies between expertise and technology.

Is there any way big models and specialized models can work together?

A persistent debate around Gen AI centers on the strategic selection between the utilization of expansive, pre-trained models like GPT and the pursuit of a more specialized path via the fine-tuning of models using proprietary data. According to Matt, an ideal outcome would be to harmonize both the approaches. Broad models with large pre-trained ones can seamlessly coexist with specialized models, effectively serving as orchestrators and interpreters of their outcomes. This hybrid methodology ensures a balance between comprehensive knowledge and task-specific expertise, helping organizations stay adaptable and navigate diverse scenarios effectively.

Data in its old or underlying form can be a mess, in what ways can Gen AI help?

The rise of Gen AI also necessitates a re-examination of data strategies embedded within organizations. The core objective must be to establish data infrastructure that not only possesses robustness but also ensures accessibility for domain experts. As the barriers between highly technical work and innovative ideas fall, data management will adopt a forward-looking approach - one that recognizes the evolving centrality of data as a foundation to Gen AI-powered success. And while there could be different problems with the underlying data, Gen AI excels at handling messy data in a way that other tools find it impossible to. It not only handles it with ease but can map old data into new data formats and can look for anomalies and missing values.

Finally, is it here to stay or something of a fad?

In the life sciences industry, Gen AI isn't just a passing trend; it's a genuine paradigm shift. Organizations must shed their reservations and grasp the opportunity to harness its monumental transformative potential. The barriers are falling, and innovation is on the horizon. Those who enthusiastically embrace it will find themselves at the forefront of what is sure to be an innovative and competitive future.

Related news

Show more

Follow us


View more