In recent years there has been a dramatic increase in the number of times the concept of AI has been mentioned during discussions at conferences, as technology has developed and the public have become more aware of it. At almost every conference Pharmora has attended, there has been a mention of AI.  

Bionow decided that AI should get its own conference, which one of our scientists attended, eager to hear the industry insights about the advancements and uses of AI in the Bioscience and Biotech industries.  

The day started off with a fascinating talk from one of the sponsors for this event, AstraZeneca, describing how they are trying to develop a Graph Neural Network (GNN) to look for alternate uses for some of their products, perhaps even in different therapy areas. GNNs are not true AIs, but a very complex algorithm, which looks at complex data and creates a map of interconnections and communications between different nodes. This could be used to look at signalling pathways or interactions between molecules in cells.  

Should this GNN be successfully developed, it could theoretically be a big money- and time-saver for the company, and good news for patients too. However, this is still in development, and our attention was drawn to the limitations of this GNN. This GNN is working with a huge amount of data, and with great complexity. They found issues in scalability and over-smoothing. Human validation at this stage would always been required, as there is not 100% confidence in this GNN yet. Humans would also have to be heavily involved in answering why a product might work in a different therapy area.   

Panel discussion included the diagnosis of 2 types of companies regarding AI:  

• Debating ethics and incredibly wary 

• Buying every AI tool and service available 

The panel also mentioned that they had mostly seen a big uptake of AI in Sales and Marketing, and Manufacturing, rather than in research and clinical development. Though, one panel member did mention an increase in the use of AI in clinical trial design and recruitment of patients.  

The panel then went on to talk about the possible problems with AI, and where they could see some things going wrong.  

Firstly, AI is already being actively used, and there is normally not any disclosure when it is. An example was given of regulatory documents being written using AI. Taking that a step further, if an AI is then developed that reviews regulatory documents, we could get to a stage where no human looks at a regulatory document before it is approved. Only after approval, when the first humans start to read it, could errors or hallucinations be spotted.  

In summary, the BioAI conference was a fascinating opportunity to take a look into what the opinions of industry leaders are on AI. The general consensus in the room was that although there is a great potential for AI to support the bioscience industry, regulations will have to be put in place to ensure that AI is being used safely and sensibly.  

Please follow this link to visit our main website and find out more about Pharmora!