🏥 AI in the NHS Newsletter #11
9th-16th August 2025
📊 Executive Summary
This week witnessed vibrant debates about AI's dual impact on NHS workflows—from revolutionary time-saving tools transforming GP appraisals to the unintended consequences of AI-generated job applications flooding recruitment systems. The community explored critical questions about clinical deskilling, the economics of AI scribes, and the stark regulatory differences between global healthcare markets, whilst celebrating small victories in personal productivity and sharing candid insights about implementation challenges.
🎯 Major Discussion Themes
1. The AI Recruitment Paradox: When Efficiency Backfires 📝
The group uncovered a fascinating paradox in NHS recruitment. Multiple members shared firsthand experiences of AI's unintended consequences in hiring:
From the frontlines: A GP partner revealed finding ChatGPT prompts left in salaried GP applications—a moment of dark comedy that sparked serious discussion. Another member's spouse, recruiting doctors for their Trust, reported receiving "20 near-identical letters that just regurgitate the job description at greater length."
The NIHR perspective highlighted the broader impact: "They are receiving unprecedented numbers of applications that they used to be able to immediately discount, but now they all sound legit and they have to sit down and properly review all." The knock-on effect? Six-month waiting times for responses, with interviews still a month away—severely impacting research timelines.
The procurement parallel: A seasoned voice noted how this mirrors NHS procurement, where "bigger players" use a "shiny brochure approach" specifically designed to exclude smaller, local innovators. As one member succinctly put it: "Terrible for the buyer, terrible for innovation."
Solutions emerged: One recruitment expert suggested that competent employers will need to "extend their shortlist pool and interview more people to evade the AI disguised inappropriate candidates," whilst others advocated for asking candidates to reveal their prompts or using AI detection tools.
2. Revolutionary Personal Productivity: The Plaud Success Story 🎙️
In stark contrast to recruitment woes, one member's experience with Plaud (an AI voice recorder) became the week's most celebrated success story:
"Custom Plaud template based on COSTAR format and GP tools criteria for appraisal. Vented about my past year into plaud after dropping kids to a party, and 5 mins later a cogent, detailed and verbose summary of my past year in practice... What was a 2 hour job done in 25 mins. I love my PLAUD!"
The group enthusiastically endorsed this approach, with the newsletter curator adding: "Venting and also answering questionnaires and forms with Plaud is ace." This sparked discussions about GPT-5 integration in Plaud and the broader potential for voice-based documentation tools.
3. The Deskilling Debate: Learning from Aviation ✈️
The Lancet's warning about clinical deskilling through AI reliance prompted sophisticated discussion. One member drew parallels with aviation: "In US, airline pilots are required to fly planes manually for a proportion of flight time to maintain cognitive sharpness."
Historical perspective added depth: "Time was doctors were great at tasting urine," one member quipped, highlighting how medical practice has always evolved with technology. Another noted the irony: "CT has replaced clinical examination for many years."
The consensus: Whilst acknowledging technology's benefits, the group emphasised maintaining core competencies. As one radiologist observed: "Every time there's a new promise of freeing up time, it just gets swallowed up by having more scans to do."
4. Global AI Arms Race: India vs UK Regulatory Approaches 🌍
A fascinating comparison emerged between international approaches to clinical AI:
The Indian example: A podcast discussion revealed an Indian hospital network going "all in on LLM within EHR" with full access to patient data, labs, imaging, and guidelines, providing continuous copilot guidance. "By all accounts very successful," but would attract "very high risk classification in UK/EU (to the point of being pretty much impossible right now)."
The regulatory dilemma: This highlighted the tension between innovation and safety, with one member noting: "Some regions are tearing ahead in the Clinical AI arms race," whilst acknowledging "multiple hazards in this setup."
5. The Economics of AI Scribes: ROI Reality Check 💰
A sobering discussion with a US hospital chain CEO revealed uncomfortable truths about AI scribe economics:
"They see the value, but they feel cost is prohibitive & it only releases clinical Pajama Time & not increased productivity. So more cost to the system but no improvement in productivity."
The morale factor: Multiple members expressed frustration that "improved staff QoL and morale is not valued," with one noting the hidden costs of burnout: decreased productivity, poorer outcomes, increased litigation, and staff churn.
Market dynamics: Encouragingly, scribes are "heading towards free-for-all in the US," with companies like Doximity offering free tools. Epic's upcoming integrated scribe tool may further disrupt the market.
📈 Group Statistics & Engagement
Activity Metrics
Total messages: 187 messages across the week
Peak activity: Friday 16th August (morning surge 07:00-12:00)
Quietest period: Weekend of 10th-11th August
Average daily messages: 27 messages
Top Contributors
The Newsletter Curator (18 messages) - Bridging discussions, sharing resources
The Procurement Expert (15 messages) - Reality checks on NHS systems
The Clinical Safety Voice (14 messages) - Balancing innovation with caution
The US Healthcare Connector (13 messages) - International perspectives
The Radiology Duo (12 messages combined) - Ground-truth from imaging
Hottest Topics by Engagement
AI in recruitment (42 messages, 8+ contributors)
Plaud/scribe experiences (31 messages, passionate advocacy)
Clinical deskilling concerns (28 messages, philosophical depth)
Global regulatory differences (19 messages, eye-opening comparisons)
GPT-5 performance debate (15 messages, technical analysis)
😄 Lighter Moments & Group Dynamics
Comedy gold: "I had applications for a salaried GP post where they had left the prompts in the responses..." sparked both laughter and despair.
The zimmer frame moment: "I had to explain what tactile vocal fremitus was to a GP registrar just recently. I walked off on my zimmer shaking my head."
Culinary medical education: Discussion of beetroot-stained urine led to a wider conversation about squid ink pasta's effects—"useful to know in the bougie populations!"
AI fails: A relative's attempt to animate 1970s family photos resulted in AI interpreting a small monkey on someone's shoulder as a Black woman—"One guess which platform... rhymes with blok."
Leaving the country: When someone mentioned their excellent colleague emigrating to Australia, another quipped: "What did you do that led to them all leaving the country?!?" Response: "We didn't have any incriminating photos of her to stop her leaving. A complete failure on our part..."
💬 Quote Wall
"All this will mean is that competent and good employers will have to extend their shortlist pool and interview more people to evade the AI disguised inappropriate candidates."
"That improved staff QoL and morale is not valued is pretty damning IMHO"
"Some regions are tearing ahead in the Clinical AI arms race"
"What was a 2 hour job done in 25 mins... I love my PLAUD!"
"CT has replaced clinical examination for many years 😁"
"Terrible for the buyer, terrible for innovation..."
"We should learn from airline industry on how to prevent de-skilling"
"Just when one ceiling to scaling emerges, another window opens"
📚 Journal Watch
Academic Papers & Key Articles
1. Bias in AI Medical Imaging
📎 Springer - Journal of Medical Systems
New evidence of baked-in bias in medical AI systems. This paper examines how algorithmic bias persists in medical imaging applications, with implications for health equity and the need for diverse training datasets.
2. The Lancet on Clinical Deskilling
📎 The Register Coverage
The Lancet raises significant concerns about healthcare professionals becoming deskilled through over-reliance on AI tools. The article argues for maintaining core clinical competencies whilst adopting new technologies.
3. Homer Simpson Effect in Medical Genetics
📎 BMJ 2016
A cautionary reminder about genetic privacy—referenced when a member jokingly warned about divulging genetic profiles. The paper discusses implications of genetic data sharing and privacy concerns.
4. NHS Strike Impact Data
📎 NHS England Report
Official NHS data showing record treatment numbers in June despite industrial action. Provides context for discussions about system capacity and efficiency.
5. AI in Robotics - World Humanoid Robot Games
📎 The Guardian
Coverage of China's advances in humanoid robotics, highlighting both breakthroughs and current limitations in the technology.
Industry & News Articles
6. Automated Job Applications Crisis
📎 The Telegraph
Investigative piece on how unemployed individuals are using AI chatbots to mass-apply for positions, directly relevant to the group's recruitment discussions.
7. Doximity's Free AI Scribe
📎 Fierce Healthcare
Major market development as Doximity enters the AI scribe market with a free offering, potentially disrupting the economics discussed by the group.
8. UK Worker Satisfaction Study
📎 The Telegraph
Comparative study showing UK workers report lower satisfaction than counterparts in India and Philippines, providing context for discussions about staff morale and burnout.
9. Windows 12 AI Integration
📎 Windows Central
Microsoft's plans for voice-first UI in Windows 12, sparking concerns about open-plan office practicality and accessibility issues.
Technical Resources
10. ProRL v2 - Scaling AI Training
📎 Notion Document
Technical documentation on new approaches to AI scaling, suggesting "just when one ceiling to scaling emerges, another window opens."
11. NHS GDE Medical Device Integration Blueprint
📎 NHS Future Platform
Practical guidance on integrating medical devices with new EPR systems, shared as a valuable resource for implementation teams.
12. New Zealand's Position on Generative AI
📎 Te Whatu Ora
New Zealand's health service position on LLMs, noted as being more progressive than current UK stance, highlighting international regulatory variations.
🔮 Looking Ahead
Upcoming Events
Clinical Jam: Sunday 28th September at Wilmslow Health Centre
Portfolio Career Workshop 3: Early February 2026, likely Manchester
St Andrews Health Centre Session: October/November (date TBC)
Discord AKT Revision Server: Now live for GP trainees
Unresolved Discussions
Central NHS position on generative AI (awaiting policy beyond June CCIO letter)
Insurance and indemnity for AI-assisted clinical decisions
Standardisation of medical coding for AI summarisation
ROI methodology for tools that improve morale but not "productivity"
Emerging Themes
Free AI scribe tools entering the market
Epic's integrated scribe launch imminent
Windows 12's voice-first UI concerns
Need for vendor-independent AI assessment framework
🎭 Group Personality Snapshot
This week perfectly captured the group's unique character: unflinchingly honest about challenges whilst celebrating genuine breakthroughs. The community's strength lies in its ability to blend frontline frustration with strategic thinking, technical expertise with human insight, and serious professional discourse with perfectly-timed humour.
The simultaneous discussions of AI transforming personal productivity whilst creating systemic recruitment problems exemplifies the group's nuanced understanding that technological progress isn't uniformly distributed—it solves some problems whilst creating entirely new ones.
What makes this community special is its refusal to accept simple narratives. When discussing clinical deskilling, members don't just worry—they propose solutions from other industries. When confronting regulatory barriers, they don't just complain—they examine what's working elsewhere. And when someone leaves ChatGPT prompts in their job application, they don't just mock—they recognise it as a symptom of deeper systemic issues.
Newsletter compiled from 187 messages between 9th-16th August 2025. Join us in continuing these vital conversations about AI's role in transforming healthcare.
Next Issue: Issue 12 will capture another week of insights from the frontline of NHS AI implementation.
Remember: "Just when one ceiling to scaling emerges, another window opens"—whether that's in AI development or NHS transformation.
