On The Ground At BPI: Where AI Dreams Meet Operational Reality

By Tyler Menichiello, Chief Editor, Bioprocess Online

A lot of this year’s BioProcess International (BPI) program was centered around AI and digitalization. While there’s no doubt that AI will remain a hot talking point across conference agendas for the foreseeable future, it’s nice to see the conversation finally moving away from the hype and more towards the actual progress being made. (Read my recent article on AstraZeneca’s use of AI in its shifts towards predictive process control.)

While it’s encouraging to hear about case studies and the lessons learned, there is still a disconnect between the promises of AI and the current realities of biomanufacturing. Bridging this divide over the coming years will be critical to realizing the next generation of biopharma production.

James Colley, Ph.D., senior director of BioPhorum’s IT and Digital Data Phorum, addressed this challenge in his BPI presentation titled “Closing the Gap Between AI Ambition and Operational Reality in Pharma.”

Bioprocess Online was fortunate enough to connect with Colley ahead of his session for an interview on the state of AI adoption in biopharma.

The following transcript has been edited for clarity and readability.

You’re giving a talk on closing the AI gap between ambition and operational reality in pharma. Would you mind giving our audience a bit of insight as to what you hope to touch on in and what you hope the audience takes away from your talk?

Colley: As a Phorum director in BioPhorum, I get the perspective from a lot of different senior executives in the biopharmaceutical industry. We hear a lot about their challenges, what they're working on, where they're trying to go, and their roadmaps. I feel what I can do at this conference is reflect that back to the audience and give a very frank appraisal of where we are with AI in the industry.

To contextualize this gap, where would you say we are in terms of AI adoption? In your abstract, you said only about four or so companies have integrated AI in a true sense. Where do you see the biggest disconnect between where we want to be and where we are right now?

Colley: There are a few challenges; one is how rapidly this is all moving. Even now, when I think about what I wrote in the session abstract, that's moved on; to claim it's four companies now is probably not true anymore. I've worked in digital, transformational technologies before. Nothing has moved as fast as this or has changed as fast as this, and that is a permanent challenge for the industry.

On an operational and technical level, we've been talking about data for years, and we'll continue talking about it. What I’m encouraged by is that this has really changed the business's perspective on data. I think there's a focus there, which we've been asking for as an industry for years, but I think it's really happening now.

Where do you hope to see the biggest change moving forward to get us to that fully autonomous point where teams are not only reactive to changes in the process, but can have predictive control?

Colley: This is very much of interest to BioPhorum and the IT Digital and Data Phorum in particular. BioPhorum’s digital plant maturity model is something that our members often use to understand where they are now in their digital journey, as well as to plan those roadmaps for moving forward through predictive plants and into adaptive plants, and that long-term view of Industry 4.0 or lights-out manufacturing, whatever you want to call it.

There are lots of things going on there.

I don't want to harp on too much about the data, but the data is a priority. I think another place a lot of work is happening at the moment is around agentic AI. A lot of investment is going into reducing the amount of time it takes to do something. So, we’re often looking at months to minutes now on technologies. We're looking to change something by tenfold — not 10% or 50%. We're looking for absolute, radical change through the adoption and the implementation of AI.

What is one piece of advice you would give to our audience as it relates to adopting AI moving forward into this next-gen biomanufacturing landscape?

Colley: That's a really tough question. What we're hearing from our members now, and what I think we're probably all appreciating is, there's so much AI now. There's so much on offer. I'm hearing our members say, “You can't digest it all.”

You really have to try and understand which AI solutions are going to be important for your role and focus on those as something to build expertise in. We're still hearing from members that human-in-the loop is absolutely critical, but more than that, the expertise that this human holds is still critical. We’re not offloading all of our expertise and knowledge into AI; we’re using AI to support those people and to support their sector expertise. That's still really important for us.

A third one that I think is linked to that is, what does the digitally native workforce of the future look like? How are we going to have people who retain expertise if we're offloading a lot of this into different areas? The more we rely on agentic AI to help us with our knowledge, where are the next generation of scientists and engineers going to come from who've got that experience and who've been hands-on in labs and in manufacturing facilities?

We're creating a conundrum that we're going to have to figure out around the digital workforce. It's something that touches on me as well. In addition to being a senior forum director, I'm also a father of two kids. I'm also thinking about their future. I think a lot about what I hear in the press and the importance of how we’re going to adapt to use AI.

BioPhorum is also really interested in environmental sustainability. I'd say one of my concerns at the moment is thinking about the environmental impact of the data centers and how the carbon burden of that is going to come through and affect the carbon burden of our products. There's a lot more to consider than just how we’re going to apply this to our technology.