Google Cloud exec sees 2025 as a tipping point for GenAI ROI in life sciences

The irony isn’t lost on Google’s leadership: even the architects of artificial intelligence find themselves navigating its transformative effects. “You know, things are never quiet,” remarked Shweta Maniar, global director of healthcare and life sciences at Google Cloud in a recent interview, recalling how Silicon Valley giants vowed to disrupt healthcare among other industries. “I remember being on the receiving end of that, thinking: nobody in any industry—but particularly in this industry—nobody wants to be disrupted, right?” she said.

Yet here we are, as the rapid ascent of genAI now challenges even those at the forefront. “Things are moving so fast that we, as a tech company, are being disrupted,” she added. With her background spanning biotech, medical devices, and healthcare advisory roles, Maniar doesn’t foresee a loss of AI momentum.

While enterprise genAI adoption varies across industries, overall adoption continues to build. According to a McKinsey survey published on May 30, 2024, 65% of respondents report that their organizations were regularly using genAI—nearly double the percentage from just ten months ago. Major tech companies continue to invest significant sums in developing AI services and infrastructure while Goldman Sachs noted on September 18 that tech earnings were outperforming the global market while AI investments have steadily increased across software and hardware/semiconductor firms.

For innovators in the life sciences ecosystem, the stakes are high, and the risks of taking a ‘wait-and-see’ approach are real. Elsevier’s “Insights 2024: Attitudes toward AI” report found that 95% of surveyed researchers expect AI to accelerate knowledge discovery. Some 87% anticipate improved work quality, and 85% think AI will free them up for higher-value research. Yet the survey, involving just under 3,000 respondents in late 2023 and early 2024, revealed that only 31% of researchers were using AI, while 67% expect to adopt it in 2–5 years. The difference between thriving and falling behind may hinge on how quickly organizations can navigate the genAI learning curve.

The 2025 tipping point: genAI ROI set to materialize for many

If 2023 and 2024 were about genAI experimentation, 2025 will likely see the tech coming into its own, driving efficiency in drug discovery and clinical trials, and helping organizations tailor treatments more precisely. “One key trend we’re seeing across the industry is, organizations want to be able to identify the right treatment for the right patient at the right time,” Maniar said. “We no longer accept that it often takes 10 to 20 years for personalized medicine solutions to come to market.…”

Reuters recently reported that the executive attendees at its NEXT conference forecast that autonomous agents and profitability will dominate the AI agenda in 2025, thanks in part to continued progress in AI reasoning.

By 2025, more projects that began as small pilot programs in prior years will scale. McKinsey’s findings indicate business units across several industries adopting genAI often see both cost decreases and revenue jumps. In the life sciences arena, genAI is seeing growing use for tasks ranging from administrative tasks like literature review and report summarization to complex drug development simulations.

The Deloitte report titled “Realizing transformative value from AI and Generative AI in life sciences” describes how a top 10 biopharma company with average revenue of $65-75 billion could capture between $5–7 billion in peak value by scaling AI over five years. The firm expects the value creation to come from three main areas: research and development (30–45% of value), commercial applications (25–35%), and manufacturing and supply chain improvements (15–25%). The report emphasizes that organizations should start with “no regrets bets” — high-value, low-complexity initiatives like scientific literature summarization and automated clinical study report authoring — to build momentum and fund further AI investments down the line. Multiple well-designed genAI deployed in concert can increase the likelihood of substantive organizational gains. “2025 is going to be where we’re going to start to see scale and some pretty significant ROI derived in the health and life science industry,” she predicted.

Weaving together genAI threads

The transformative potential of genAI emerges from integrated deployment across the value chain. “We’re seeing natural language processing for analyzing data not only in early research but in manufacturing practices, supply chain, and the commercial side of bringing medicines to market,” Maniar said.

GenAI is making its mark where it can create the most value. The aforementioned McKinsey survey found that organizations, across industries, most often deploy genAI in marketing and sales and in product and service development. For life sciences, these insights translate into more efficient internal data retrieval, improved patient engagement, and streamlined R&D. “Think about if you are working for a large pharmaceutical organization—when you are looking for something in your internal corpus of information, now, because of genAI, you can get personalized information based on what you’re searching for,” Maniar said.

Crawl, walk, run: Building trust in AI agents

Mature adoption of AI will be a gradual process, with human oversight remaining a crucial consideration. One place this dynamic is in sharp relief is with the rise of AI agents, a key trend Maniar identifies. Starting with routine tasks like data entry and literature review, these agents will gradually take on more complex responsibilities as trust builds. “You have to remember that a lot of the folks who will be using AI agents are not historically AI practitioners,” she noted. “So building trust into these systems is going to show its usage in areas where there’s a lot of historical information and data that you can compare.…”

While AI agents can handle routine tasks, trust and understanding must be earned. “It’s not just AI practitioners who must trust these systems. We need transparency, education, and trust for everyone. We’re all going to be users of this technology, so we need to build trust and understanding industry-wide,” Maniar said.

Over time, as trust grows, AI agents can take on more complex tasks freeing employees to gradually shift away from tedium to focus more on high-value activities.

From needing more data to finding patterns in the data you already have

AI’s ability to rapidly process large, diverse datasets reveals actionable patterns faster than ever. “A good example is how we’ve worked with Ginkgo Bioworks on AI search for efficient data access,” Maniar noted. This exemplifies two key trends she sees driving adoption: multimodal AI and intuitive search. By integrating text, images, and genomic data, companies can quickly uncover insights previously hidden in fragmented repositories. The goal isn’t just collecting more data, but extracting new value from existing information. As Maniar put it, “We are moving away from this world of needing more data… It’s ‘How do I elicit new patterns from the existing information and data that I have?’”

A spectrum of AI advances—from AlphaFold’s protein modeling to AI-driven agents accelerating research—now enables research teams to focus on deeper scientific questions. With growing emphasis on finding new patterns in existing data, the life sciences industry appears ready for meaningful gains. The potential gradual rise of AI agents could hasten this transformation. “I think what this allows us to do is reframe what we use our skilled teams for,” Maniar explains. “We can offload repetitive and tedious tasks to these AI agents. That’s going to be the big shift. Ideally, that will allow us to leverage the unique capabilities we, as people, have in the life science industry.” The result, she says, will be “shifting focus to higher-value skilled work.”