In the face of declining productivity and increasing R&D costs, biopharmaceutical companies—industry wide—are searching for ways to become more effective and more efficient. Embracing artificial intelligence (AI) is a compelling option. But, like companies in many other industries, although biopharma leaders and personnel may be psychologically ready to accept AI, they aren’t necessarily prepared for that integration.
Integration is constrained by a few factors. Cultural buy-in is still occurring, workers need to be trained to integrate effectively into their work, and some elements of the data and computing infrastructures need to be refined or added, points out Alistair Henry, PhD, executive vice president and CSO at UCB, a global biopharmaceutical company based in Brussels. Consequently, he says, “We have readiness at one level, but perhaps not preparedness.”
Any new technology has curious, technology-savvy people who become early adopters, “but they don’t drive cultural change. That has to come from the top,” continues Henry. Therefore, it behooves senior leadership to set the expectation that the organization will use new technologies to enhance what they do.
People who can integrate AI into their roles effectively—bridging gaps between, for example, biologists and computational scientists—are invaluable. [Andresr/Getty Images]
Regulatory acceptance is also a concern. Typically, regulatory acceptance of new technologies lags behind technological advances, and the AI space is still being charted. That said, “Regulators are there to ensure safety and efficacy. Anything we can do that strengthens the data package to give clarity and enable them to assess the product appropriately…they will embrace,” Henry stresses. “They are embracing AI like everybody else, but (also like everybody else) they’re not sure what they’re embracing.”
Cultural acceptance is still a hurdle, and people throughout an organization are concerned about how generative AI will change their jobs. Some wonder whether their roles will exist a few years from now. “AI is an enabler, not a replacer,” notes Henry. “AI is not replacing bench scientists.”
Even post-doctoral positions, he says, should be safe. Why? Because, “I expect my post-docs to think. I still want smart people to think about the biology and take the inputs from the generative process to look at the big picture.” In that environment, AI is an enhancement that helps bench scientists become more efficient.
AI is changing the skills needed to succeed, however. “A digitally-literate workforce is super critical,” emphasizes Henry.
Skillsets also need to be fluid to keep up with the steadily accelerating pace of technological innovation. Even though individuals may not be able to write the algorithm or understand exactly how an AI application reaches its conclusions, they need to understand both the value and potential risks that AI usage brings to the organization. People who can integrate AI into their roles effectively—bridging gaps between, for example, biologists and computational scientists—are invaluable.
But no one is expected to master AI and their own field. “That’s why there are teams,” he says, and those interdisciplinary teams bring value. “Through those collaborations, you find things you didn’t expect, because people from different disciplines ask different questions that spark different conversations,” according to Henry. “That’s incredibly important.”
AI is a tool. It’s therefore vital to upskill employees throughout the organization to use AI effectively, so they understand the benefits (such as generating a comprehensive analysis of disparate data sets that shed light on multiple projects) as well as AI’s limitations. “It’s trained on what you think you know,” says Henry.
Workforce training is the piece that is largely missing in organizations today, he adds, and is what they need to prioritize. Preparing for the current digital transformation, therefore, should focus around “training, education, culture, and leadership.”
Currently, AI adoption seems to be related to organizational size. Large and mid-sized biopharmaceutical organizations are integrating AI into deeper and deeper levels of operations already, while smaller biotechs may have used AI in their discovery programs and are working to find additional benefits of this technology.
“There is no one-size-fits-all for AI,” explains Henry. “AI isn’t like a mobile phone. It is a computational methodology that you have to apply in different ways in different types of data sets. There are lots of different AI versions. We talk about machine learning, but there are also all sorts of different computational models that have different demands and complexities that are used across the whole value chain.”
Alistair Henry, PhD
Rather than attempting to develop AI applications in-house, “Look for alliances with AI companies,” Henry advises. Some AI-enabled applications are available off-the-shelf, but specialty programs are more typically the result of collaborations. “The technology is moving so fast that companies need partnerships. The trick is finding groups that are developing methodologies and building infrastructure around specific questions.”
In this AI-enabled world, data is a commodity. Any one data set now has value across projects and across disciplines, and throughout a product lifecycle, from molecule generation, through hypothesis generation, clinical trial design, and manufacturing. Therefore, datasets must be organized so they can be searched easily, used to address many different questions, and be widely accessed enterprise wide.
As Henry states, “AI lets you use data, you otherwise may have left on the table, to inform other programs.” While he can’t share the details publicly, he says making this change at UCB “has dramatically changed our choices and the number of molecules we made.”
He cites AI’s value in helping researchers prioritize which molecules to make. Because it can access and analyze all the data rather than only subsets, scientists can identify the most promising molecular candidates and eliminate red herrings by performing more meaningful experiments sooner. This reduces costs and time to trials.
When coupled with automation, AI can help manufacturing become more efficient, too. As an example, Henry suggests using AI to evaluate the usage of reagents to identify how they may be used more efficiently or even recycled, thus reducing costs as well as environmental impacts.
It may be counterintuitive, but a plethora of data can actually hinder innovation. That is what happened when it first became possible to accumulate what came to be called “big data.” When inundated with data, possible options increased exponentially and innovators became overwhelmed by possibilities and, often, unable to see and make connections within the data.
“AI,” he says, “can help them prioritize, based on the data.” In practical terms, for example, AI can help researchers select the five most promising experiments from a possible 50, thereby eliminating the need to perform 45 experiments, and the concomitant time and costs. “Restricting resources drives thinking, innovation, and creativity,” Henry asserts. “Not everybody shares this view, but by restricting resources, people are forced to prioritize, so they must think (about projects) more critically.”
AI isn’t meant to replace critical thinking, he stresses. It is meant to eliminate the noise to help people think critically about the elements that matter. To do that, people throughout the organization need to learn how to use AI effectively.
The post Biopharma is Mentally Ready, Yet Unprepared for AI Integration appeared first on GEN - Genetic Engineering and Biotechnology News.
Integration is constrained by a few factors. Cultural buy-in is still occurring, workers need to be trained to integrate effectively into their work, and some elements of the data and computing infrastructures need to be refined or added, points out Alistair Henry, PhD, executive vice president and CSO at UCB, a global biopharmaceutical company based in Brussels. Consequently, he says, “We have readiness at one level, but perhaps not preparedness.”
AI adoption roadblocks
Any new technology has curious, technology-savvy people who become early adopters, “but they don’t drive cultural change. That has to come from the top,” continues Henry. Therefore, it behooves senior leadership to set the expectation that the organization will use new technologies to enhance what they do.
People who can integrate AI into their roles effectively—bridging gaps between, for example, biologists and computational scientists—are invaluable. [Andresr/Getty Images]
Regulatory acceptance is also a concern. Typically, regulatory acceptance of new technologies lags behind technological advances, and the AI space is still being charted. That said, “Regulators are there to ensure safety and efficacy. Anything we can do that strengthens the data package to give clarity and enable them to assess the product appropriately…they will embrace,” Henry stresses. “They are embracing AI like everybody else, but (also like everybody else) they’re not sure what they’re embracing.”
Cultural acceptance is still a hurdle, and people throughout an organization are concerned about how generative AI will change their jobs. Some wonder whether their roles will exist a few years from now. “AI is an enabler, not a replacer,” notes Henry. “AI is not replacing bench scientists.”
Even post-doctoral positions, he says, should be safe. Why? Because, “I expect my post-docs to think. I still want smart people to think about the biology and take the inputs from the generative process to look at the big picture.” In that environment, AI is an enhancement that helps bench scientists become more efficient.
A digital workforce
AI is changing the skills needed to succeed, however. “A digitally-literate workforce is super critical,” emphasizes Henry.
Skillsets also need to be fluid to keep up with the steadily accelerating pace of technological innovation. Even though individuals may not be able to write the algorithm or understand exactly how an AI application reaches its conclusions, they need to understand both the value and potential risks that AI usage brings to the organization. People who can integrate AI into their roles effectively—bridging gaps between, for example, biologists and computational scientists—are invaluable.
But no one is expected to master AI and their own field. “That’s why there are teams,” he says, and those interdisciplinary teams bring value. “Through those collaborations, you find things you didn’t expect, because people from different disciplines ask different questions that spark different conversations,” according to Henry. “That’s incredibly important.”
AI is a tool. It’s therefore vital to upskill employees throughout the organization to use AI effectively, so they understand the benefits (such as generating a comprehensive analysis of disparate data sets that shed light on multiple projects) as well as AI’s limitations. “It’s trained on what you think you know,” says Henry.
Workforce training is the piece that is largely missing in organizations today, he adds, and is what they need to prioritize. Preparing for the current digital transformation, therefore, should focus around “training, education, culture, and leadership.”
Currently, AI adoption seems to be related to organizational size. Large and mid-sized biopharmaceutical organizations are integrating AI into deeper and deeper levels of operations already, while smaller biotechs may have used AI in their discovery programs and are working to find additional benefits of this technology.
One-size-doesn’t-fit-all
“There is no one-size-fits-all for AI,” explains Henry. “AI isn’t like a mobile phone. It is a computational methodology that you have to apply in different ways in different types of data sets. There are lots of different AI versions. We talk about machine learning, but there are also all sorts of different computational models that have different demands and complexities that are used across the whole value chain.”
Alistair Henry, PhD
Rather than attempting to develop AI applications in-house, “Look for alliances with AI companies,” Henry advises. Some AI-enabled applications are available off-the-shelf, but specialty programs are more typically the result of collaborations. “The technology is moving so fast that companies need partnerships. The trick is finding groups that are developing methodologies and building infrastructure around specific questions.”
In this AI-enabled world, data is a commodity. Any one data set now has value across projects and across disciplines, and throughout a product lifecycle, from molecule generation, through hypothesis generation, clinical trial design, and manufacturing. Therefore, datasets must be organized so they can be searched easily, used to address many different questions, and be widely accessed enterprise wide.
As Henry states, “AI lets you use data, you otherwise may have left on the table, to inform other programs.” While he can’t share the details publicly, he says making this change at UCB “has dramatically changed our choices and the number of molecules we made.”
He cites AI’s value in helping researchers prioritize which molecules to make. Because it can access and analyze all the data rather than only subsets, scientists can identify the most promising molecular candidates and eliminate red herrings by performing more meaningful experiments sooner. This reduces costs and time to trials.
When coupled with automation, AI can help manufacturing become more efficient, too. As an example, Henry suggests using AI to evaluate the usage of reagents to identify how they may be used more efficiently or even recycled, thus reducing costs as well as environmental impacts.
It may be counterintuitive, but a plethora of data can actually hinder innovation. That is what happened when it first became possible to accumulate what came to be called “big data.” When inundated with data, possible options increased exponentially and innovators became overwhelmed by possibilities and, often, unable to see and make connections within the data.
“AI,” he says, “can help them prioritize, based on the data.” In practical terms, for example, AI can help researchers select the five most promising experiments from a possible 50, thereby eliminating the need to perform 45 experiments, and the concomitant time and costs. “Restricting resources drives thinking, innovation, and creativity,” Henry asserts. “Not everybody shares this view, but by restricting resources, people are forced to prioritize, so they must think (about projects) more critically.”
AI isn’t meant to replace critical thinking, he stresses. It is meant to eliminate the noise to help people think critically about the elements that matter. To do that, people throughout the organization need to learn how to use AI effectively.
The post Biopharma is Mentally Ready, Yet Unprepared for AI Integration appeared first on GEN - Genetic Engineering and Biotechnology News.