The Reality of AI: Q&A with Shweta Maniar

As global director of healthcare and sciences at Google Cloud, Shweta Maniar is working with organizations to find ways to implement AI into their workflows. She spoke with Pharmaceutical Executive about the state of AI and how companies are experimenting with it.

Pharmaceutical Executive: How is AI being implemented to automate documentation?
Shweta Maniar: Generative AI (gen AI) is starting to fundamentally transform regulatory processes by helping automate documentation and acting as a quality control expert. It can automate the creation of documents like clinical trial reports and manufacturing protocols by drafting text from existing data (such as patient information or lab results) and summarizing complex scientific information.

This automation can reduce manual effort and minimize the potential for human error. Additionally, gen AI serves as a powerful quality control tool by automatically identifying language inconsistencies, flagging potential issues, and ensuring all documents adhere to a consistent voice and format. This is transformative, as regulatory reviews have historically been a manual and error-prone process.

PE: A common complaint about gen AI is that the work still needs to be double-checked due to hallucinations and other errors. Is this still an issue and how can it be prevented?
Maniar: While hallucinations and errors in gen AI outputs can occur, Google Cloud is taking a deliberate approach to address this. For example, we’re releasing new gen AI offerings with built-in guardrails that undergo rigorous internal and external testing to ensure they meet user needs and high safety standards. In addition, products like Google Cloud’s Vertex AI Search use only an organization’s data and that organization can “ground” gen AI outputs to this data, which helps reduce the risks of hallucinations or inaccurate responses by citing and linking to original, internal sources.

Companies need to have clean, organized data to make AI tools truly useful and for the FDA and other regulatory bodies to effectively use AI-powered reviews. The shift towards a "digital-first" mindset, where information is structured for computer-skimmability, helps mitigate these issues by providing AI with high-quality, consistent inputs.

In fact, AI's strength lies in its ability to manage the massive volume and complexity of regulatory submissions, which can involve analyzing lengthy documents created by dozens or even hundreds of individuals. And by using it to detect inconsistencies and flag potential issues before human review, AI can help maintain a consistent tone and style across vast amounts of information, reducing the likelihood of errors that might otherwise require extensive fact-checking.

That said, keeping people in the loop of this new AI landscape isn't just a feature—it's fundamental. For example, we build our AI to be as transparent as possible with citations to backup all outputs, so that experts can always see what’s going on under the hood. It’s this ability to check the work that builds real trust in the system.

PE: What timeline can the industry expect for AI to deliver on the promise of reduced responses from FDA?
Maniar: While we’re at the beginning of this industry transformation, the regulatory landscape is shifting toward much faster feedback cycles with the use of AI and other technologies. Given the rapid pace of AI development, we can expect AI to significantly support reviews and feedback in the near term.

My advice is for organizations to prepare for this change immediately. The era of receiving feedback in days, not months, is on the horizon. The benefits are substantial for both patients, who will gain faster access to life-saving therapies, and for companies, which can prevent millions of dollars in lost revenue for every day a drug is delayed.

PE: Is it too early to start thinking about shifting workflows?
Maniar: It's a critical imperative to start shifting workflows now. The question is no longer if AI will change regulatory submissions, but how your organization will evolve as regulatory bodies do. Regulatory agencies around the globe are already adopting AI tools, signaling a fundamental shift.

To keep pace, companies must embrace a digital-first strategy, beginning with preparing their data for AI consumption and incorporating AI into team workflows. This also demands a cultural shift, where organizations must train employees to use AI agents to streamline tasks like identifying relevant information and automating multi-step processes.

This transition transforms the regulatory process from a protracted "black hole" of waiting into a dynamic and iterative cycle. Ultimately, this ensures more rapid market access and delivers critical innovations to patients sooner. Companies that prepare now won't just keep pace—they'll set it, turning regulatory hurdles into a competitive advantage and delivering therapies to patients faster than ever before.

The Reality of AI: Q&A with Shweta Maniar

Key Takeaways

Related Stories

Newsletter