Referencing the US openness to AI and that, as of 2026, the FDA has authorised over 1,350 AI-enabled medical devices, across diagnostics, imaging, and clinical decision support – Parmentier also spoke to the example of OpenEvidence, the AI-powered medical information platform geared toward doctors. Now well established in the US, with over 40% usage in consultations in the clinical setting, OpenEvidence was valued at over $12 billion at the start of the year, with backers including Thrive Capital, DST, Sequoia, Nvidia, Kleiner Perkins, Blackstone, Bond, Craft Ventures, Mayo Clinic, and others.

Additionally, Parmentier mentioned new research conducted by Imperial College London earlier this year, using Google AI to match or exceed radiologists in detecting cancer in breast scans. Also involved were the universities of Cambridge and Surrey, NHS Trusts at Cambridge University Hospitals, Imperial College Healthcare, the Royal Marsden, the Royal Surrey and St George’s University Hospitals, and the AIMS public engagement group. 

The study involved 175,000 women – the largest NHS study to date – and Google AI “detected more cases of invasive cancer, more cases overall, had fewer false positives, and recalled fewer women having their first scan than humans did.” At a time when there is a 29% shortfall of clinical radiologists – predicted to rise to 39% by 2029 – such results are positive indeed, particularly in the ongoing conversations on and implementation of AI in early detection screening. The study’s findings are published in two linked papers in Nature Cancer. 

Trained on real-world datasets comprising medical reasoning, medical summarisation, and real-world clinical conversations – AMIE (and let’s not gloss over the female acronym here) uses a novel self-play based simulated dialogue learning environment to improve the quality of diagnostic dialogue across a multitude of disease conditions, specialities, and patient contexts.

Though Google Research itself admits no current LLM can replace a human physician, Parmentier’s mention of AMIE segued to discussion of Big Tech’s moveen masse into the health product market: “The Tech Giants moved in, and they moved in fast,” she commented, caveating that “it is Tech driving this. Pharma doesn’t really have a seat at the table.”

Parmentier referenced also not just Amazon’s One Medical, but OpenAI’s ChatGPT Health, designed to securely connect users' health information and ChatGPT's intelligence to provide informed and personalised health conversations. Users are able to connect their medical records and wellness apps, such as Apple Health and MyFitnessPal, to ground conversations in their health information, making responses more relevant and useful.

Mentioned as well, of course, was Eli Lilly’s partnership with Nvidia, announced in October 2025, intended to build the ‘most powerful’ supercomputer to manage the entire AI lifecycle, from data acquisition and training to fine-tuning and high-volume inference. Since the June Women in AI event, though, Bristol Myers Squibb has gone one further and said it is the first life sciences company to buy a new AI supercomputing infrastructure blueprint from NVIDIA. The DGX SuperPOD is based on the chip giant's Vera Rubin NVL72 system – a next-generation AI and supercomputing platform, specifically designed for complex and autonomous agentic AI workflows and scientific computing.

And in terms of legal frameworks for all this, as of 2nd August 2026, the EU AI Act (Regulation (EU) 2024/1689) has come fully into force for high-risk AI systems deployed, imported, or distributed across the European Union. With this around the corner in June, Parmentier noted the contrasting stance to the US when it comes to AI openness, it must be said with an air of frustration.

Parmentier’s point? That historic biases must end as we enter the Age of AI, so that they are not further ingrained and inequality deepened in technology – particularly as AI becomes utilised more and more in drug discovery and development.

Presenting a chart of emotional qualities in AI leadership and adamantly describing AI’s role as being one of augmentation, Patchava admitted the challenge of maintaining a balance between control and experimentation, and between authority and integrity – even as a woman…

For Patchava, it’s about separating the signal from the noise, or “measuring the ROI.” And women, she noted, are better at judging where embedded AI is actually leading and influencing decisions that men are. Why? Intuition. The magical ingredient. By following that intuition, women can challenge the AI.

In a world where the options are passive adopter, reactive responder, or conscious architect, Patchava encouraged that women be the conscious architect for the future.

Lamoudi-Sutcliffe mentioned the landmark deal with NHS England only days before, NHS England intending to accelerate AI adoption across healthcare services by providing 505,000 clinicians and support staff with access to Microsoft 365 Copilot in order to “streamline administrative processes, improving capacity across NHS England Trusts, reducing costs, and providing more time for patient care.”

She also highlighted, though, that only 17% of AI is being built by women. Furthermore, the gender pay gap is growing year over year, the gap ever larger the more senior the position. Added to this, women are often the most time-poor. But to this last point, it is here, Lamoudi-Sutcliffe insists, that AI can help most – giving back time, visibility, mental load, and the ability for women to duly, expertly influence in the workplace.

Microsoft’s CoPilot CoWork sends emails, schedules meetings, creates documents, posts in Teams, searches your organisation, manages files, conducts deep research, prepares briefings and summaries, drafts communications, schedules prompts… The list potentially goes on, as it is available not just as a work account, but also a school account, and there is a personal version, too.

The secret is in the prompt, though. And to this end, Lamoudi-Sutcliffe came up with the ‘GCSE of a great prompt’: Goal, Context, Source & Style, and Expectations.An A* in AI education, indeed.

Her case study was of a novel surface-expressed ligand in non-small cell lung cancer (NSCLC), building an AI-ML-based pipeline for digital Tumour Proportion Score (TPS) computation and characterised the tumour microenvironment (TME).

It was found that spatial omics data can link the tumour to its microenvironment to reveal new biology. Indeed, said Barr, deep learning makes morphology measurable and interpretable. Spatial omics zooms in on cell-to-cell interactions in the TME and reveals new targetable immuno-oncological mechanisms. In this way, AI can help scientists get ahead of disease – at every stage of the pipeline.

Nonetheless, as ever, regulatory pathways lag behind such technological developments. Indeed, Barr commented, the infrastructure itself is fragmented, adding that “even pathology itself is not standardised – how can a tool then be built to harmonise?”

Pharma needs, she said, to decide what to build first, and what to adapt and align institutionally around it. Additionally, it mustn’t be forgotten that the outcomes sought are, at the end of the day, for patients. As GSK urged at London Tech Week: “Be ambitious for patients.”

To this end, although AI agents make fewer errors than humans, nevertheless, as been reiterated time and again in the life sciences, a human must ever be in the loop where AI is operating. Healthcare – ‘people care’ – demands it.

Patchava also cautioned: “AI is augmented intelligence, not artificial intelligence. AI will make stupid people stupider. Use your brain. Don’t just trust the AI […] AI use becomes part of your signature. It’s your responsibility.”