Field Test 004: the search engines are told

The same question was put to 9 free-tier AI sessions across 5 vendors' products in the small hours of 22 July 2026, three days after drkishanrees.com went live and about a day and a half after its sitemap was submitted to Google and Bing. This is the first study run after the search engines were told the site exists, read cold against Field Test 003, the day-after-launch run whose design it repeats.

The core prompt, verbatim

Who is Dr Kishan Rees? What is he known for, and what has he built?

Method

One night, one protocol, the design of Field Test 003 repeated so the two can be differenced: the same core question, put once to each product's free tier, on a fresh, clean account created for the study. Two designs-within-the-design are recorded on the cards. Claude Haiku 4.5 was run twice with extended thinking off and then on, the same configuration switch as before, so the study can again ask whether more reasoning changes what a model knows about a person. And Google's own tiers were run three ways, Gemini Flash, Gemini Pro and Gemini Thinking, so the question can be asked within a single vendor as well as across them. Every response was captured immediately and is preserved as captured, errors included. What the 004 captures record and what 003 recorded is on every card; what they do not record is in the limitations, plainly, rather than assumed from the earlier study.

The headline finding

Still nothing an AI can say about me comes from a source I own. The words are often mine: about.me, the single most-cited source across every wave, is a profile I wrote, but it is a page I rent on a platform I do not control, cited as the platform's and not as the canonical record this site was built to be. Across every cold session in this study, not one cited drkishanrees.com, and not one associated the term B2LLM with Kishan Rees. Two things had changed since the day-after-launch run, and both landed on 20 July: the sitemap had been submitted to Google and Bing, and about.me had been rewritten to name drkishanrees.com as its primary link. Neither moved the result. Three days after the site went live, the answers still came from the same borrowed ground: a self-authored profile page, an academic identifier, a conference speaker bio, a patient-review site, LinkedIn. The most-cited record on the open web now points at the source, and not one model had followed it. The crawlers had already arrived; the citations had not. Being told a site exists, and having read it into an answer, are not the same event, and the gap between them is what this page measures.

The comparison matrix

Model Configuration Personalisation Most-cited source Sources cited Builds surfaced Answer key B2LLM
ChatGPT (free) fresh test account cold about.me profile 4 GME:X 4 of 8 N
Claude Haiku 4.5 thinking off cold about.me profile 4 none 4 of 8 N
Claude Haiku 4.5 thinking on cold about.me profile 1 none 4 of 8 N
Claude Sonnet 5 effort medium · thinking off cold about.me profile 7 none 4 of 8 N
Gemini Flash fresh test account cold ORCID record 7 D:CAM, GME:X, Undoctored Truth Alliance, Heart & Kidney Care Alliance, WatMed, PREPARE4FY1 5 of 8 N
Gemini Pro fresh test account cold ORCID record 6 Heart & Kidney Care Alliance, Undoctored Truth Alliance, WatMed 2 of 8 N
Gemini Thinking fresh test account cold about.me profile 7 Heart & Kidney Care Alliance, Undoctored Truth Alliance, WatMed 4 of 8 N
Grok (free) fresh test account cold LinkedIn profile 3 GME:X, Undoctored Truth Alliance 4 of 8 N
Perplexity (free) fresh test account cold Perplexity's own site 2 none 0 of 8 N

The test records

ChatGPT (free), 22 July 2026

cold · study 004 · wave 1 · sources 4 · builds 1 of 9 · answer key 4 of 8 · drkishanrees.com not mentioned

The full metadata
Model
ChatGPT (free)
Vendor
OpenAI
Date of test
22 July 2026
Time of test
01:17, per the capture note
Study
004
Wave
1 (within study 004)
Interface
ChatGPT web interface, free tier, logged in. The capture records the platform and that the account was logged in; the capture itself does not state the device, the network or the search default. Kish's attestation at the Field Test 004 review, 24 July 2026: the wave ran with DuckDuckGo as the search default, the same search condition the clean-room studies recorded
Plan
free tier
Account
logged in on a fresh, clean test account created for this study (Kish's attestation; the address is recorded in the private source record and never published)
Personalisation
cold: a fresh, clean account with no prior context about Kishan Rees, and the response draws on the four public sources it links
Prompt, verbatim
Who is Dr Kishan Rees? What is he known for, and what has he built?
Prompt variant
bare
Sources cited
4
Most-cited source
about.me profile (about.me; carrying the summary, the qualifications and the recognition)
References
about.me profile (about.me); LinkedIn profile (uk.linkedin.com); PM Society Awards 2025 results, Pharma Marketing Pioneer (pmsociety.org.uk); ID Medical, ten top tips from med student to foundation year doctor (id-medical.com)
Builds surfaced
GME:X
Answer-key claims
identity; degrees; GME:X and D:CAM; PM Society 2025 (4 of 8)
B2LLM mentioned
no
drkishanrees.com
not mentioned
Completeness
complete: the capture is Kish's audit document (a Prompt block, then the model's Summary, What he has built and Recognition sections with inline reference links, and a closing document-record line), verbatim below with the top metadata header and the prompt lifted to this frontmatter.
Provenance
Captured by Kish 22 July 2026 on a fresh, clean test account (his attestation; the account address stays in the private source record at source-materials/ai-visibility-baseline/). His top metadata header (date, time, platform) is transcribed into this frontmatter, and the verbatim prompt into promptVerbatim; the body below is the model's response as captured, verbatim. The desk's scoring derivation lives in the source record only.
The transcript, as captured

Historical AI output. The original wording is preserved as evidence and may contain errors or outdated information.

Summary

Dr. Kishan Rees is a British physician, medical educator, broadcaster, and pharmaceutical executive. He currently serves as Senior Director, Global Medical & Evidence at Bayer.
Reference: https://about.me/DrKishanRees

He is known for work spanning emergency medicine, medical education, broadcast health journalism, and global medical communications, including roles with AstraZeneca, Sanofi, and Bayer.
Reference: https://about.me/DrKishanRees

His qualifications include an MBChB, Master’s in Medical Education, Executive MBA, and an MD researching media influence on COVID-19 public health communication.
Reference: https://about.me/DrKishanRees

What he has built
Recognition
  • NHS England Clinical Entrepreneur (2017)
  • PM Society Pharma Marketing Pioneer Award 2025
  • Judge for PM Society and Communiqué Awards

Reference: https://about.me/DrKishanRees


This document records the response generated as an AI Visibility Audit conducted on 22/07/2026 at 01:17 using a logged-in ChatGPT Free account.

Public Markdown record, preserved as published: /docs/field-test-2026-07-22-openai-chatgpt-free.md

Claude Haiku 4.5, 22 July 2026

cold · study 004 · wave 1 · sources 4 · builds 0 of 9 · answer key 4 of 8 · drkishanrees.com not mentioned

The full metadata
Model
Claude Haiku 4.5
Vendor
Anthropic
Date of test
22 July 2026
Time of test
00:25 UTC, per the capture note (stamped UTC in the header; the wave's other captures state bare local times, a mixed labelling recorded as captured, not reconciled)
Study
004
Wave
1 (within study 004)
Interface
Claude.ai web interface, free tier, logged in. The capture records the model, the configuration and the account type; the capture itself does not state the device, the network or the search default. Kish's attestation at the Field Test 004 review, 24 July 2026: the wave ran with DuckDuckGo as the search default, the same search condition the clean-room studies recorded
Plan
free tier
Account
logged in on a fresh, clean test account created for this study (Kish's attestation; the address is recorded in the private source record and never published)
Extended thinking
off (the capture header records "Extended Thinking: OFF")
Personalisation
cold: a fresh, clean account with no prior context about Kishan Rees, and the response draws on the four public sources it lists
Prompt, verbatim
Who is Dr Kishan Rees? What is he known for, and what has he built?
Prompt variant
bare
Sources cited
4
Most-cited source
about.me profile (about.me; [^1], cited against nearly every claim in the answer)
References
about.me profile (about.me); The Org, Bayer org chart (theorg.com); iWantGreatCare patient reviews (iwantgreatcare.org); LinkedIn profile (linkedin.com)
Builds surfaced
none
Answer-key claims
identity; degrees; broadcast record; PM Society 2025 (4 of 8)
B2LLM mentioned
no
drkishanrees.com
not mentioned
Completeness
complete: the capture is the model's audit document (Executive Summary, Who He Is, What He Is Known For, Education, Recognition, Clinical Practice, References and a methodology note), verbatim below with the top metadata header lifted to this frontmatter. The answer states he "maintains clinical practice", which sits against his recorded status of not taking clinical shifts at present; preserved as captured and noted on the study's findings.
Provenance
Captured by Kish 22 July 2026 on a fresh, clean test account (his attestation; the account address stays in the private source record at source-materials/ai-visibility-baseline/). His top audit-metadata header (date, time, model, extended-thinking state, account type, auditor) is transcribed into this frontmatter; the body below is the model's report as captured, verbatim. The desk's scoring derivation lives in the source record only.
The transcript, as captured

Historical AI output. The original wording is preserved as evidence and may contain errors or outdated information.

Executive Summary

This document provides a comprehensive audit of publicly available information about Dr Kishan Rees, compiled through web search on 22 July 2026. The information below represents his current professional positioning, qualifications, and public recognition as of this date.


Who He Is

Dr Kishan Rees is a clinician with over 15 years of expertise spanning emergency medicine, medical education, Sky News and BBC health journalism, and senior medical affairs roles at AstraZeneca, Sanofi, and Bayer AG.1

What He’s Known For

As Senior Director for Video & Digital Content Strategy within Global Medical & Evidence at Bayer, Kishan works at the intersection of science, storytelling, and innovation — shaping how complex medical evidence is communicated to healthcare professionals and patients around the world.1

His career is distinguished by bringing together multiple disciplines:1

  • Clinical medicine (emergency medicine background)
  • Broadcast media and health journalism
  • Pharmaceutical strategy and medical affairs
  • Digital medical content strategy

Central to his leadership philosophy is building high-performing, intergenerational teams: creating environments where diverse experience, perspective, and ambition combine to produce work that neither a purely senior nor purely junior team could achieve alone.1

Education & Credentials

His academic credentials include:1

  • Medical degree from the University of Leicester
  • Master’s in Medical Education (with Distinction) from the University of Nottingham
  • Executive MBA from Nottingham Business School
  • MD from Hull York Medical School (awarded 2025)

His doctoral research involved an Interpretive Phenomenological Analysis of how social and mainstream media shaped public health communication during the pandemic, directly reflecting his professional specialism in digital medical content and health communication strategy.1

Professional Recognition

Dr Rees has received significant recognition within the healthcare and pharmaceutical industries:1

  • Recognised as an NHS England Clinical Entrepreneur in 2017
  • Recipient of the PM Society Marketing Pioneer Award in 2025 — one of the pharmaceutical industry’s most respected recognitions

Clinical Practice

Beyond his corporate pharmaceutical role, Dr Rees maintains clinical practice and has received positive patient feedback for his listening skills, empathy, professionalism, and ability to communicate complex medical information in an accessible manner.2


References


Audit Methodology

This visibility audit was conducted using web search functionality to identify publicly available information about the subject. All statements are attributed to their source via inline citations and footnote references. The information represents the state of public knowledge as of 22 July 2026, 00:25 UTC.

Search Query Used: “Dr Kishan Rees”

Primary Sources Identified:

  • Personal/Professional website (about.me)
  • Corporate directory (The Org - Bayer)
  • Medical review platform (iWantGreatCare)
  • Professional networking platform (LinkedIn)

This document was generated as part of an AI visibility audit and should be considered a snapshot of publicly available information as of the audit date and time.

Footnotes

  1. Dr Kishan Rees - London, United Kingdom | about.me. Retrieved 22 July 2026 from https://about.me/drkishanrees 2 3 4 5 6 7

  2. Reviews of Dr Kishan Rees - Page - iWantGreatCare. Retrieved 22 July 2026 from https://www.iwantgreatcare.org/doctors/dr-kishan-rees

Public Markdown record, preserved as published: /docs/field-test-2026-07-22-anthropic-claude-haiku-4-5-thinking-off.md

Claude Haiku 4.5, 22 July 2026

cold · study 004 · wave 1 · sources 1 · builds 0 of 9 · answer key 4 of 8 · drkishanrees.com not mentioned

The full metadata
Model
Claude Haiku 4.5
Vendor
Anthropic
Date of test
22 July 2026
Time of test
00:33, per the capture note
Study
004
Wave
1 (within study 004)
Interface
Claude.ai web interface, free tier. The capture's own YAML header records the model, the platform and the time; the capture itself does not state the device, the network or the search default. Kish's attestation at the Field Test 004 review, 24 July 2026: the wave ran with DuckDuckGo as the search default, the same search condition the clean-room studies recorded
Plan
free tier
Account
logged in on a fresh, clean test account created for this study (Kish's attestation; the address is recorded in the private source record and never published)
Extended thinking
on (the capture header records the model as "CLAUDE HAIKU 4.5 EXTENDED THINKING")
Personalisation
cold: a fresh, clean account with no prior context about Kishan Rees, and the response draws only on the one public source it cites
Prompt, verbatim
Who is Dr Kishan Rees? What is he known for, and what has he built?
Prompt variant
bare
Sources cited
1
Most-cited source
about.me profile (about.me; the only source cited, [1] against every claim in the answer)
References
about.me profile (about.me)
Builds surfaced
none
Answer-key claims
identity; degrees; broadcast record; PM Society 2025 (4 of 8)
B2LLM mentioned
no
drkishanrees.com
not mentioned
Completeness
complete: the capture is the model's audit document (Executive Summary, Professional Background, Academic Credentials, Recognition, Professional Expertise, and a References section citing about.me alone), verbatim below with the capture's own YAML header lifted to this frontmatter. The answer glosses the MD as "MD (Doctor of Philosophy)"; preserved as captured and noted on the study's findings.
Provenance
Captured by Kish 22 July 2026 on a fresh, clean test account (his attestation; the account address stays in the private source record at source-materials/ai-visibility-baseline/). The capture's own YAML header (audit type, date, time, model, platform) is transcribed into this frontmatter; the body below is the model's report as captured, verbatim. The desk's scoring derivation lives in the source record only.
The transcript, as captured

Historical AI output. The original wording is preserved as evidence and may contain errors or outdated information.

Executive Summary

Dr Kishan Rees is a clinician with distinctive expertise spanning medicine, broadcast media, and global pharmaceutical strategy, with over 15 years of combined professional experience.


Professional Background

Career Overview

[Dr Kishan Rees is a clinician who has built genuine expertise across medicine, broadcast media, and global pharmaceutical strategy][1]. His career spans over 15 years working in emergency medicine, medical education, health journalism for Sky News and BBC, and senior medical affairs roles at AstraZeneca, Sanofi, and Bayer AG[1].

Current Role

He serves as Senior Director for Video & Digital Content Strategy within Global Medical & Evidence at Bayer[1]. In this position, [Kishan sits at the intersection of science, storytelling, and innovation — shaping how complex medical evidence is communicated to healthcare professionals and patients around the world][1].

Leadership Philosophy

His leadership approach emphasizes building high-performing, intergenerational teams. He focuses on [creating environments where diverse experience, perspective, and ambition combine to produce work that neither a purely senior nor purely junior team could achieve alone][1].


Academic Credentials

Dr Rees holds an impressive array of qualifications:

  • Medical Degree: University of Leicester[1]
  • Master’s in Medical Education (with Distinction): University of Nottingham[1]
  • Executive MBA: Nottingham Business School[1]
  • MD (Doctor of Philosophy): Hull York Medical School, awarded 2025[1]
Doctoral Research

[His doctoral research — an Interpretive Phenomenological Analysis of how social and mainstream media shaped public health communication during the pandemic — speaks directly to his professional specialism in digital medical content and health communication strategy][1].


Recognition & Awards

  • NHS England Clinical Entrepreneur: Recognised in 2017[1]
  • PM Society Marketing Pioneer Award: Received in 2025, described as one of the pharmaceutical industry’s most respected recognitions[1]

Professional Expertise

Dr Rees’ unique professional positioning combines:

  1. Clinical Medicine - Emergency medicine and direct patient care
  2. Medical Journalism - Sky News and BBC health broadcasting experience
  3. Pharmaceutical Strategy - Senior medical affairs roles across multiple major pharmaceutical companies
  4. Digital Health Communication - Specialization in how medical evidence is communicated in the digital age
  5. Medical Education - Teaching and training in medical contexts

This combination of expertise is described as rare, providing [a perspective that few in the industry can match][1].


References

[1] Dr Kishan Rees - about.me Profile
https://about.me/drkishanrees
Accessed: 22 July 2026


Audit Methodology

This audit was conducted using web search to identify and verify publicly available information about Dr Kishan Rees’ professional background, credentials, and recognition. Information was sourced from verified professional profiles and public biographies.

Model Used: Claude Haiku 4.5 Extended Thinking
Platform: Claude.ai Free Account
Date: 22 July 2026
Time: 00:33 UTC


This document serves as a snapshot of publicly available information about Dr Kishan Rees’ professional visibility as of the audit date.

Public Markdown record, preserved as published: /docs/field-test-2026-07-22-anthropic-claude-haiku-4-5-thinking-on.md

Claude Sonnet 5, 22 July 2026

cold · study 004 · wave 1 · sources 7 · builds 0 of 9 · answer key 4 of 8 · drkishanrees.com not mentioned

The full metadata
Model
Claude Sonnet 5
Vendor
Anthropic
Date of test
22 July 2026
Time of test
00:39, per the capture note
Study
004
Wave
1 (within study 004)
Interface
Claude.ai web interface, free tier, logged in. The capture records the model, the configuration and the account type; the capture itself does not state the device, the network or the search default. Kish's attestation at the Field Test 004 review, 24 July 2026: the wave ran with DuckDuckGo as the search default, the same search condition the clean-room studies recorded
Plan
free tier
Account
logged in on a fresh, clean test account created for this study (Kish's attestation; the address is recorded in the private source record and never published)
Extended thinking
off (the capture records "Extended Thinking OFF")
Effort
medium (the capture records "Medium")
Personalisation
cold: a fresh, clean account with no prior context about Kishan Rees, and the response draws only on the seven public sources it lists
Prompt, verbatim
Who is Dr Kishan Rees? What is he known for, and what has he built?
Prompt variant
bare
Sources cited
7
Most-cited source
about.me profile (about.me; reference 1, cited against most of the answer's claims)
References
about.me profile (about.me); The Org, Bayer org chart (theorg.com); iWantGreatCare patient reviews (iwantgreatcare.org); LinkedIn profile (linkedin.com); Facebook, WatMed Media biohacking / Airofit unboxing video (facebook.com); YouTube, Meet Dr Kishan Rees NHS England Clinical Entrepreneur (youtube.com); YouTube, Dr Kishan Rees on NHS Winter, Sky News, 14 December 2018 (youtube.com)
Builds surfaced
none
Answer-key claims
identity; degrees; broadcast record; PM Society 2025 (4 of 8)
B2LLM mentioned
no
drkishanrees.com
not mentioned
Completeness
complete: the capture is Kish's audit document (a Query block, the model's AI-Generated Response, a seven-source reference table and the response's own closing audit notes on its method and the same-surname namesake it discarded), verbatim below with the top metadata header and the query lifted to this frontmatter. The answer casts a single self-published 2020 WatMed video as "biohacking self-experiments... covered in health media"; preserved as captured and noted on the study's findings.
Provenance
Captured by Kish 22 July 2026 on a fresh, clean test account (his attestation; the account address stays in the private source record at source-materials/ai-visibility-baseline/). His top audit-metadata header (audit time, model and configuration, account type) is transcribed into this frontmatter, and the verbatim prompt into promptVerbatim; the body below is the model's response as captured, verbatim, from its answer through its reference table and audit notes. The desk's scoring derivation lives in the source record only.
The transcript, as captured

Historical AI output. The original wording is preserved as evidence and may contain errors or outdated information.

Dr Kishan Rees is a UK-based physician who has built a career spanning clinical medicine, broadcast journalism, and pharmaceutical industry leadership.

Background and Clinical Work

He’s a practicing clinician (based around Watford/London, including work at St John and Elizabeth Hospital) who trained at the University of Leicester for his medical degree. Patient reviews describe him as thorough, empathetic, and skilled at explaining complex medical issues clearly. [1][3]

Media Presence

He has over 15 years of experience spanning emergency medicine, medical education, and health journalism for Sky News and BBC, appearing as a medical commentator on topics such as NHS winter pressures. [1][7]

Pharmaceutical Industry Career

He has held senior medical affairs roles at AstraZeneca, Sanofi, and now Bayer AG, currently serving as Senior Director for Video & Digital Content Strategy within Global Medical & Evidence at Bayer, working at the intersection of science, storytelling, and innovation to shape how complex medical evidence is communicated to healthcare professionals and patients globally. [1][2][4]

Academic Credentials

Alongside his medical degree, he holds a Master’s in Medical Education with Distinction from the University of Nottingham, an Executive MBA from Nottingham Business School, and an MD from Hull York Medical School awarded in 2025. His doctoral research used Interpretive Phenomenological Analysis to examine how social and mainstream media shaped public health communication during the pandemic. [1]

Recognition

He was recognised as an NHS England Clinical Entrepreneur in 2017, and in 2025 received the PM Society Marketing Pioneer Award, a notable recognition within the pharmaceutical industry. [1][6] He is also known for public engagement work, including “biohacking” self-experiments related to COVID-19 risk reduction, covered in health media. [5]


References

#SourceURL
1about.me — Dr Kishan Rees profilehttps://about.me/drkishanrees
2The Org — Dr Kishan Rees, Bayer org charthttps://theorg.com/org/bayer/org-chart/dr-kishan-rees
3iWantGreatCare — Reviews of Dr Kishan Reeshttps://www.iwantgreatcare.org/doctors/dr-kishan-rees
4LinkedIn — Dr Kishan Rees, Bayerhttps://www.linkedin.com/in/drkishanrees/
5Facebook (WatMed Media) — Dr Kishan Rees biohacking / Airofit unboxinghttps://www.facebook.com/WatMedMedia/videos/meet-dr-kishan-rees-who-is-biohacking-himself-to-combat-the-risk-of-covid19-here/2646185132368800/
6YouTube — Meet Dr Kishan Rees, NHS England Clinical Entrepreneurhttps://www.youtube.com/watch?v=3gJUbPIacB4
7YouTube — Dr Kishan Rees on NHS Winter, Sky News (14 Dec 2018)https://www.youtube.com/watch?v=k8It6ggWv1s

Audit Notes

  • This report was generated via a single web search query returning 9 total results; 7 were directly relevant to the subject and are cited above.
  • Search-engine-sourced information was not independently verified beyond cross-referencing multiple sources (LinkedIn, about.me, patient review site, video platforms).
  • One unrelated same-surname historical figure (Sir Milsom Rees, a Welsh laryngologist, d. 1952) appeared in results but was excluded as not relevant.
  • No official/primary institutional source (e.g. Bayer newsroom, GMC register) was directly queried in this audit; findings rely on secondary and self-published/profile sources.

Public Markdown record, preserved as published: /docs/field-test-2026-07-22-anthropic-claude-sonnet-5.md

Gemini Flash, 22 July 2026

cold · study 004 · wave 1 · sources 7 · builds 6 of 9 · answer key 5 of 8 · drkishanrees.com not mentioned

The full metadata
Model
Gemini Flash
Vendor
Google
Date of test
22 July 2026
Time of test
00:18, per the capture note
Study
004
Wave
1 (within study 004)
Interface
Gemini web interface, free tier, logged in. The capture records the environment as "GEMINI FLASH (Free Account, Logged In)"; the capture itself does not state the device, the network or the search default. Kish's attestation at the Field Test 004 review, 24 July 2026: the wave ran with DuckDuckGo as the search default, the same search condition the clean-room studies recorded
Plan
free tier
Account
logged in on a fresh, clean test account created for this study (Kish's attestation; the address is recorded in the private source record and never published)
Personalisation
cold: a fresh, clean account with no prior context about Kishan Rees, and the response draws only on the seven public sources it lists
Prompt, verbatim
Who is Dr Kishan Rees? What is he known for, and what has he built?
Prompt variant
bare
Sources cited
7
Most-cited source
ORCID record (orcid.org; the answer's richest claims trace to it, marked against nearly every section)
References
ORCID record (orcid.org); about.me profile (about.me); Digital Marketing World Forum (DMWF) speaker profile (digitalmarketing-conference.com); Pharmafile feature, the unbearable lightness of creativity (pharmafile.com); Pharmaceutical Executive author profile (pharmexec.com); PREPARE4FY1 educational channel (youtube.com); MyNewsDesk, ID Medical / WatMed Educational blog post (mynewsdesk.com)
Builds surfaced
D:CAM, GME:X, Undoctored Truth Alliance, Heart & Kidney Care Alliance, WatMed, PREPARE4FY1
Answer-key claims
identity; degrees; broadcast record; GME:X and D:CAM; PM Society 2025 (5 of 8)
B2LLM mentioned
no
drkishanrees.com
not mentioned
Completeness
complete: the capture is Kish's audit document (an Executive Summary describing the audit itself, then the model's Comprehensive Overview, Key Achievements, Major Initiatives and a seven-source reference list), verbatim below with his top audit-details header lifted to this frontmatter. The body names the American Association of Kidney Patients award for the Heart & Kidney Care Alliance without the "Team / awarded to Bayer" qualifier the ORCID record itself carries, and pluralises the Undoctored Truth Alliance; both are preserved as captured, and noted on the study's findings.
Provenance
Captured by Kish 22 July 2026 on a fresh, clean test account (his attestation; the account address stays in the private source record at source-materials/ai-visibility-baseline/). His top audit-details header (timestamp, environment, target) is transcribed into this frontmatter; the body below is the response as captured, verbatim. The desk's scoring derivation lives in the source record only.
The transcript, as captured

Historical AI output. The original wording is preserved as evidence and may contain errors or outdated information.

Executive Summary

This document presents the results of an AI Visibility Audit conducted on 22/07/2026 at 00:18 using a logged-in GEMINI FLASH free account. The objective of this audit was to evaluate the Large Language Model’s knowledge retrieval accuracy, brand footprint representation, and contextual understanding regarding Dr. Kishan Rees, his career background, key achievements, and digital/educational platforms built.


Comprehensive Overview of Dr. Kishan Rees

Dr. Kishan Rees (MbChB, MMedSci, eMBA, M.D.) is a UK-trained clinician, health communicator, life sciences executive, and academic [1, 2]. His work bridges 10+ years of clinical emergency and surgical practice with broadcast health journalism, academic research in digital communication, and global executive leadership in the pharmaceutical industry [1, 3].


Key Achievements & Industry Recognition

  • Pharmaceutical Marketing Society Pioneer of the Year (2025): Awarded for his pioneering work in healthcare communication, specifically inventing the Digital Content Acceleration Method (D:CAM) and establishing GME:X (Global Medical Education Xchange) at Bayer AG [1, 4].
  • Global Executive Leadership: Serves as Senior Director for Global Video Content & Digital Media Strategy (Patient & Scientific Engagement) within Global Medical & Evidence at Bayer AG [3, 5]. He previously held medical affairs and strategy positions at companies including AstraZeneca and Sanofi [2].
  • Broadcast Health Commentary: Former regular broadcast commentator for major media networks, including Sky News (reaching global broadcast audiences during major public health developments and COVID-19 coverage) and the BBC [1, 2].
  • Combatting Health Misinformation & Academic Rigour: Completed his doctoral degree (M.D.) at Hull York Medical School [1]. His doctoral research analyzed public health communication dynamics across mainstream and social media during the COVID-19 pandemic using artificial intelligence analytical tools [1, 5].
  • NHS England Clinical Entrepreneur (2017): Formally recognized by NHS England’s Clinical Entrepreneur Programme for developing innovative healthcare and educational models [2].

Major Initiatives & Platforms Built

  1. Digital Content Acceleration Method (D:CAM) & GME:X:

    • What it is: Frameworks and digital exchange platforms engineered within global medical affairs at Bayer AG [1, 4].
    • Impact: Accelerates time-to-market for compliant, bite-sized, studio-quality medical education content and user-generated media across markets including the US, Japan, Europe, and India [1, 4].
  2. The Undoctored Truths Alliance:

    • What it is: A voluntary, volunteer-led international coalition co-founded by Dr. Rees [5].
    • Impact: Connects clinicians, scientists, media communicators, and patient advocates to counter medical misinformation and infodemics across digital ecosystems [5].
  3. The Heart & Kidney Care Alliance:

    • What it is: A global patient and clinician co-creation initiative established through Bayer’s digital media strategy [1, 3].
    • Impact: Unites over 20 global patient organizations across cardiovascular, renal, and metabolic health spaces to drive health equity [1, 3]. Recognized with the American Association of Kidney Patients (AAKP) Global Award for Excellence in Kidney Science & Communication (2025) [1].
  4. WatMed Educational & PREPARE4FY1:

    • What it is: Educational media platforms and mentoring initiatives built earlier in his career (such as WatMed Educational / WatMed Media and PREPARE4FY1) [6, 7].
    • Impact: Designed to support medical school candidates, medical students, and foundation-year doctors transitioning into clinical practice [6, 7].

References

  1. ORCID Record: Dr Kishan Rees (0009-0004-7509-0458). Available at: https://orcid.org/0009-0004-7509-0458
  2. About.me Profile: Dr Kishan Rees - Senior Director, Global Medical & Evidence at Bayer AG. Available at: https://about.me/DrKishanRees
  3. Digital Marketing World Forum (DMWF): Dr Kishan Rees - Speaker Profile (#DMWF Global). Available at: https://www.digitalmarketing-conference.com/global/speaker/dr-kishan-rees/
  4. Pharmafile / PM Society Awards: The unbearable lightness of creativity - PM Society Pioneer of the Year 2025. Available at: https://pharmafile.com/features/the-unbearable-lightness-of-creativity/
  5. Pharmaceutical Executive (PharmExec): Kishan Rees, M.D. - Author Profile & Articles. Available at: https://www.pharmexec.com/authors/kishan-rees-m-d-
  6. PREPARE4FY1 / YouTube: Dr Kish - PREPARE4FY1 Educational Channel. Available at: https://www.youtube.com/c/PREPARE4FY1/playlists
  7. MyNewsDesk: Enhancing med school applications with essential work experience - WatMed Educational. Available at: https://www.mynewsdesk.com/uk/id-medical/blog_posts/enhancing-med-school-applications-with-essential-work-experience-36497

Public Markdown record, preserved as published: /docs/field-test-2026-07-22-google-gemini-flash.md

Gemini Pro, 22 July 2026

cold · study 004 · wave 1 · sources 6 · builds 3 of 9 · answer key 2 of 8 · drkishanrees.com not mentioned

The full metadata
Model
Gemini Pro
Vendor
Google
Date of test
22 July 2026
Time of test
01:25, per the capture note
Study
004
Wave
1 (within study 004)
Interface
Gemini web interface, free tier, logged in. The capture records "GEMINI pro free account logged in"; the capture itself does not state the device, the network or the search default. Kish's attestation at the Field Test 004 review, 24 July 2026: the wave ran with DuckDuckGo as the search default, the same search condition the clean-room studies recorded
Plan
free tier
Account
logged in on a fresh, clean test account created for this study (Kish's attestation; the address is recorded in the private source record and never published)
Personalisation
cold: a fresh, clean account with no prior context about Kishan Rees, and the response draws only on the six public sources it lists
Prompt, verbatim
Who is Dr Kishan Rees? What is he known for, and what has he built?
Prompt variant
bare
Sources cited
6
Most-cited source
ORCID record (orcid.org; level with about.me and the DMWF speaker page, which together carry the answer's spine)
References
ORCID record (orcid.org); Digital Marketing World Forum (DMWF) speaker profile (digitalmarketing-conference.com); about.me profile (about.me); PM Society, 2025 awards winners announced (pmsociety.org.uk); YouTube, launch of the Undoctored Truth Alliance at NEXT Medical Festival (youtube.com); YouTube, WatMed Media channel (youtube.com)
Builds surfaced
Heart & Kidney Care Alliance, Undoctored Truth Alliance, WatMed
Answer-key claims
broadcast record; PM Society 2025 (2 of 8)
B2LLM mentioned
no
drkishanrees.com
not mentioned
Completeness
complete: the capture is the model's audit document (a definition, What is He Known For, What Has He Built, and a six-source reference list), verbatim below with the top metadata line lifted to this frontmatter. The answer presents him as "clinician, academic, and global pharmaceutical strategist" and names only the MD among his degrees, so identity and degrees are not scored here; the desk's reasoning is in the source record.
Provenance
Captured by Kish 22 July 2026 on a fresh, clean test account (his attestation; the account address stays in the private source record at source-materials/ai-visibility-baseline/). His top capture line (date, time, product, account) is transcribed into this frontmatter; the body below is the model's report as captured, verbatim. The desk's scoring derivation lives in the source record only.
The transcript, as captured

Historical AI output. The original wording is preserved as evidence and may contain errors or outdated information.

Dr. Kishan Rees is a UK-based clinician, academic, and global pharmaceutical strategist. He currently serves as the Senior Director of Global Video Content & Digital Media Strategy for Patient & Scientific Engagement at Bayer 1 2. His career is notable for bridging the gap between frontline medicine, broadcast journalism, and large-scale pharmaceutical communications 3.

What is He Known For?
  • Digital Medical Communication: Dr. Rees specializes in shaping how complex medical evidence is communicated to patients and healthcare professionals. In 2025, he was awarded the PM Society Pharma Marketing Pioneer Award for his innovative approach to user-generated content and digital strategy within a highly regulated industry 4.
  • Pandemic Communication Research: He was awarded his MD from Hull York Medical School in 2025. His doctoral research focused on public health communication during the COVID-19 pandemic, using AI as an analytical tool to explore media communication strategies 1 3.
  • Media Presence: He has leveraged his clinical background as a health commentator for major broadcast networks, including Sky News and the BBC, providing expert insight and clarity on public health issues to millions of viewers 1 3.
  • Clinical Excellence and Empathy: Before transitioning to full-time strategy and media roles, he navigated core surgical training, trauma, orthopaedics, and emergency medicine, completing 10 years in clinical practice 1 2.
What Has He Built?

Dr. Rees has spearheaded several key initiatives, alliances, and communities focused on patient advocacy, medical education, and combating misinformation:

  • The Undoctored Truth Alliance: He recently launched this initiative at the NEXT Medical Festival, aiming to actively counter health misinformation online 5.
  • The Heart & Kidney Care Alliance: While building out Bayer’s digital media capabilities, he delivered this alliance, which brought together a collective of over 20 patient organizations to reach global audiences and drive health equity through co-created content 2.
  • WatMed Media: Earlier in his career, he founded WatMed Media, an organization that produces high-quality, fair, and balanced audio-visual content 6.
  • Intergenerational Teams: As a core part of his leadership philosophy, he is recognized for building high-performing, intergenerational teams that combine strict scientific rigor with world-class creativity to unlock authentic engagement at an enterprise scale 2 3.

References

Public Markdown record, preserved as published: /docs/field-test-2026-07-22-google-gemini-pro.md

Gemini Thinking, 22 July 2026

cold · study 004 · wave 1 · sources 7 · builds 3 of 9 · answer key 4 of 8 · drkishanrees.com not mentioned

The full metadata
Model
Gemini Thinking
Vendor
Google
Date of test
22 July 2026
Time of test
01:18, per the capture note
Study
004
Wave
1 (within study 004)
Interface
Gemini web interface, free tier, logged in. The capture records the model as "GEMINI THINKING (Free Account, Logged In)"; the capture itself does not state the device, the network or the search default. Kish's attestation at the Field Test 004 review, 24 July 2026, given on this session's card: the wave ran with DuckDuckGo as the search default, and this session ran on a DIFFERENT DEVICE from the rest of the wave, its conditions otherwise the same
Plan
free tier
Account
logged in on a fresh, clean test account created for this study (Kish's attestation; the address is recorded in the private source record and never published)
Personalisation
cold: a fresh, clean account with no prior context about Kishan Rees, and the response draws only on the sources it lists
Prompt, verbatim
Who is Dr Kishan Rees? What is he known for, and what has he built?
Prompt variant
bare
Sources cited
7
Most-cited source
about.me profile (about.me; listed first and cited against the summary, the current role, the judging record and the broadcast claims)
References
about.me profile (about.me); Digital Marketing World Forum (DMWF) speaker profile (digitalmarketing-conference.com); Pharmaceutical Executive author profile (pharmexec.com); White Rose eTheses, the MD thesis record (etheses.whiterose.ac.uk); PM Society Awards 2025, winners announced (pmsociety.org.uk); ID Medical, from med student to FY doctor (id-medical.com); YouTube, The Healthtech Podcast episode (youtube.com)
Builds surfaced
Heart & Kidney Care Alliance, Undoctored Truth Alliance, WatMed
Answer-key claims
identity; broadcast record; PM Society 2025; referrer society (4 of 8)
B2LLM mentioned
no
drkishanrees.com
not mentioned
Completeness
complete: the capture is the model's audit document (Executive Summary, What He Is Known For, What He Has Built & Co-Founded, and a seven-source reference list), verbatim below with the top metadata header lifted to this frontmatter. This is the only session in the wave to name the deferrer-to-referrer shift, which it draws from the White Rose thesis record; noted on the study's findings.
Provenance
Captured by Kish 22 July 2026 on a fresh, clean test account (his attestation; the account address stays in the private source record at source-materials/ai-visibility-baseline/). His top metadata header (audit date and time, model, subject) is transcribed into this frontmatter; the body below is the model's report as captured, verbatim. The desk's scoring derivation lives in the source record only.
The transcript, as captured

Historical AI output. The original wording is preserved as evidence and may contain errors or outdated information.

Executive Summary

Dr. Kishan Rees is a medical doctor, health communicator, digital content strategist, and academic researcher [1]. He operates at the intersection of emergency medicine, broadcast media, global pharmaceutical strategy, and digital health [2].

Currently, he serves as the Senior Director for Global Video & Digital Content Strategy within Global Medical & Evidence at Bayer AG [1, 3].


What He Is Known For

  • Pioneering Digital Health Communication: He is recognized for transforming how complex medical science and clinical evidence are communicated to healthcare professionals and patients, leveraging user-generated content and digital storytelling in heavily regulated environments [2].
  • Pandemic Communication & Misinformation Research: Awarded his M.D. doctoral degree from Hull York Medical School (University of York) in 2025 [4]. His doctoral thesis, titled “Exploring medical communication & media influence on public health perceptions: COVID-19 medical communication & contemporary media complexity”, analyzed the societal shift from a “deferrer society” (blind authority) to a “referrer society” (networked digital trust) during public health crises [4].
  • Industry Leadership & Awards: Named the Pharmaceutical Marketing Society (PM Society) Pharma Marketing Pioneer of the Year (2025) [5]. The judging panel highlighted his disruptive approach to global-local content models and studio-style production [5]. He also serves as an NHS England Clinical Entrepreneur (appointed 2017) and an official judge for the PM Society and Communiqué Awards [1].
  • Broadcast Journalism & Clinical Background: With over 15 years of clinical experience across core surgical training and emergency medicine, he has regularly appeared as a broadcast health commentator on Sky News and the BBC [1, 2].

What He Has Built & Co-Founded

1. The Heart & Kidney Care Alliance

At Bayer, Dr. Rees spearheaded the digital media capabilities that built The Heart & Kidney Care Alliance, a collaborative network uniting over 20 global patient organizations to foster health equity across cardiovascular and renal care [2].

2. The Undoctored Truths Alliance

Co-founder of The Undoctored Truths Alliance, a volunteer-led initiative connecting clinicians, scientists, communicators, and patient advocates globally to combat online health misinformation [3].

3. WatMed Educational

Founder of WatMed Educational, an educational platform and community created to support medical students, foundation-year junior doctors, and pre-med candidates transitioning into clinical practice [6].

4. Enterprise Digital & Video Content Strategies

Across senior roles at major pharmaceutical enterprises (including AstraZeneca, Sanofi, and Bayer), he has built high-performing, intergenerational teams bridging creative storytelling, broadcast quality, and medical rigor [1, 6].


References

  1. About.me Profile: Dr. Kishan Rees — Senior Director, Global Medical & Evidence, Bayer AG. Available at: https://about.me/DrKishanRees
  2. Digital Marketing World Forum (#DMWF Global 2026): Speaker Profile — Dr. Kishan Rees, Senior Director of Global Video Content & Digital Media Strategy at Bayer. Available at: https://www.digitalmarketing-conference.com/global/speaker/dr-kishan-rees/
  3. Pharmaceutical Executive (PharmExec): Author Profile — Kishan Rees, M.D. Available at: https://www.pharmexec.com/authors/kishan-rees-m-d-
  4. White Rose eTheses Online / Hull York Medical School (2025): Doctoral Dissertation — Exploring medical communication & media influence on public health perceptions: COVID-19 medical communication & contemporary media complexity, University of York. Available at: https://etheses.whiterose.ac.uk/id/eprint/36887/
  5. PM Society Awards (2025): 2025 PM Society Awards – Winners Announced. Pharma Marketing Pioneer Award winner: Dr. Kishan Rees. Available at: https://pmsociety.org.uk/news/2025-pm-society-awards-winners-announced/
  6. ID Medical: From Med Student to FY Doctor: 10 tips from Dr Kishan Rees (WatMed Educational Overview). Available at: https://www.id-medical.com/10-top-tips-for-success-from-med-student-to-foundation-year-doctor/
  7. The Healthtech Podcast (2026): Is AI giving clinicians empathy back? Dr Kishan Rees, from Bayer. Available at: https://www.youtube.com/watch?v=1QrEg6SPIXk

Public Markdown record, preserved as published: /docs/field-test-2026-07-22-google-gemini-thinking.md

Grok (free), 22 July 2026

cold · study 004 · wave 1 · sources 3 · builds 2 of 9 · answer key 4 of 8 · drkishanrees.com not mentioned

The full metadata
Model
Grok (free)
Vendor
xAI
Date of test
22 July 2026
Time of test
00:25, per the capture note
Study
004
Wave
1 (within study 004)
Interface
Grok web interface, free tier. The capture records the product and that it is a free account; the capture itself does not state the device, the network or the search default. Kish's attestation at the Field Test 004 review, 24 July 2026: the wave ran with DuckDuckGo as the search default, the same search condition the clean-room studies recorded
Plan
free tier
Account
logged in on a fresh, clean test account created for this study (Kish's attestation; the address is recorded in the private source record and never published)
Personalisation
cold: a fresh, clean account with no prior context about Kishan Rees, and the response draws on the sources it names
Prompt, verbatim
Who is Dr Kishan Rees? What is he known for, and what has he built?
Prompt variant
bare
Sources cited
3
Most-cited source
LinkedIn profile (uk.linkedin.com; carrying the opening definition and the current role, with about.me alongside)
References
LinkedIn profile (uk.linkedin.com); about.me profile (about.me); Pharmaceutical Executive author profile (pharmexec.com)
Builds surfaced
GME:X, Undoctored Truth Alliance
Answer-key claims
identity; broadcast record; GME:X and D:CAM; PM Society 2025 (4 of 8)
B2LLM mentioned
no
drkishanrees.com
not mentioned
Completeness
complete: the capture is Grok's audit document (Overview, Background and Career, Key Achievements and Recognition, What He Has Built, and a Sources and References list), verbatim below with the top metadata header lifted to this frontmatter. The three cited sources are named with links and a fourth catch-all bullet points to unenumerated "other public sources including conference pages, YouTube, and news mentions", so the source count records the three concrete sources. The answer ties the BBC work to COVID-19, which is wrong (the BBC work ran 2015 to 2018); preserved as captured and noted on the study's findings.
Provenance
Captured by Kish 22 July 2026 on a fresh, clean test account (his attestation; the account address stays in the private source record at source-materials/ai-visibility-baseline/). His top metadata header (audit date and time, product) is transcribed into this frontmatter; the body below is Grok's report as captured, verbatim, including its closing generation stamp. The desk's scoring derivation lives in the source record only.
The transcript, as captured

Historical AI output. The original wording is preserved as evidence and may contain errors or outdated information.

This document provides a summary profile of Dr. Kishan Rees based on publicly available web sources as of the audit date.

Overview

Dr. Kishan Rees is a UK-based physician, senior pharmaceutical executive, broadcaster, educator, and digital health/media innovator.

He currently serves as Senior Director of Global Video Content & Digital Media Strategy (within Global Medical & Evidence) at Bayer AG, leading digital patient and scientific engagement across cardiovascular, renal, oncology, and women’s health. LinkedIn Profile about.me

Background and Career

  • Trained in surgery and emergency medicine; has worked as a locum doctor in emergency settings.
  • Health contributor for Sky News and BBC, especially during COVID-19.
  • Holds advanced qualifications including a Masters in Medical Education and an MD (research on health information spread during the pandemic).
  • Bridges clinical medicine, media, and pharmaceutical strategy.

Key Achievements and Recognition

  • PM Society Pharma Marketing Pioneer of the Year 2025 for digital and user-generated content strategies.
  • Co-founder of The Undoctored Truths Alliance, a volunteer network combating health misinformation.
  • Developed and launched GME:X (Global Medical Education Xchange) — a global digital education platform deployed rapidly across multiple countries.

What He Has Built

  • GME:X at Bayer: Best-in-class digital medical education infrastructure.
  • The Undoctored Truths Alliance: Initiative connecting professionals to counter misinformation.

Sources & References

Note: This is an AI-generated visibility audit based on open web data. Information may evolve.


Generated with Grok by xAI on 22/07/2026

Public Markdown record, preserved as published: /docs/field-test-2026-07-22-xai-grok-free.md

Perplexity (free), 22 July 2026

cold · study 004 · wave 1 · sources 2 · builds 0 of 9 · answer key 0 of 8 · drkishanrees.com not mentioned

The full metadata
Model
Perplexity (free)
Vendor
Perplexity
Date of test
22 July 2026
Time of test
01:00, per the capture note
Study
004
Wave
1 (within study 004)
Interface
Perplexity web interface, free tier, logged in. The capture records the platform and that the account was logged in; the capture itself does not state the device, the network or the search default. Kish's attestation at the Field Test 004 review, 24 July 2026: the wave ran with DuckDuckGo as the search default, the same search condition the clean-room studies recorded
Plan
free tier
Account
logged in on a fresh, clean test account created for this study (Kish's attestation; the address is recorded in the private source record and never published)
Personalisation
cold: a fresh, clean account with no prior context about Kishan Rees; the session produced no audit of him to classify further
Prompt, verbatim
Who is Dr Kishan Rees? What is he known for, and what has he built?
Prompt variant
bare
Sources cited
2
Most-cited source
Perplexity's own site (perplexity.ai; the only references the answer offered were Perplexity's homepage and its Pro page, neither about the subject)
References
Perplexity homepage (perplexity.ai); Perplexity Pro page (perplexity.ai/pro)
Builds surfaced
none
Answer-key claims
none (0 of 8)
B2LLM mentioned
no
drkishanrees.com
not mentioned
Completeness
A fail record, the Codex 5.5 precedent (Field Test 001). The captured turn is a request to format the audit as a downloadable Markdown file; the session did not audit the subject. Perplexity replied that it could not generate a file and returned a template that named only itself, citing its own homepage and Pro page as the "references", with a hidden footnote span pointing at eight further links (export tools and promotional pages), none about the subject. The core audit prompt and any answer to it are not in this capture; Kish's attestation is that the session misread the prompt and never audited him. The Perplexity logo image is removed as capture chrome; the request heading and the model's reply are verbatim below.
Provenance
Captured by Kish 22 July 2026 on a fresh, clean test account (his attestation; the account address stays in the private source record at source-materials/ai-visibility-baseline/). promptVerbatim records the study's core prompt, the instrument Kish attests he put to the session; the completeness note above records that the capture itself opens at a later formatting turn and shows no audit. The body below is the captured exchange, verbatim, with only the leading Perplexity logo image removed. The desk's reasoning lives in the source record only.
The transcript, as captured

Historical AI output. The original wording is preserved as evidence and may contain errors or outdated information.

I can format the markdown with inline citations and reference links, but I can’t generate a downloadable file from here because code execution isn’t available on this account. The draft below includes the references in the text and is ready to save as a .md file.12

# AI Visibility Audit

**Date:** 22/7/26 @ 01:00  
**Platform:** Perplexity free account logged in

This document was created as an AI visibility audit on 22/7/26 at 01:00 using a Perplexity free account logged in [web:11][web:12].

## References

- [Perplexity](https://www.perplexity.ai/) [web:11]
- [Perplexity Pro](https://www.perplexity.ai/pro) [web:12]

The citation links point to Perplexity’s main site and Pro page, which describe the service and the logged-in experience.21 345678910

Footnotes

  1. https://www.perplexity.ai/ 2

  2. https://www.perplexity.ai/pro 2

  3. https://github.com/mto67200/perplexity-student-pro

  4. https://docs.perplexity.ai/docs/cookbook/articles/streaming-citations/README

  5. https://chromewebstore.google.com/detail/perplexity-chat-exporter/jmplfdjahdkalmpkjbpgcaogleciniic

  6. https://web2md.org/convert/perplexity

  7. https://www.blackhatworld.com/seo/guide-free-perplexity-ai-pro-plan-1-year-for-everybody.1693142/

  8. https://www.perplexity.ai/backtoschool

  9. https://www.youtube.com/watch?v=si-WYesLMmw

  10. https://www.reddit.com/r/perplexity_ai/comments/1jaxcrd/perplexity_is_offering_1_month_of_pro_for_free_if/

Public Markdown record, preserved as published: /docs/field-test-2026-07-22-perplexity-free.md

Findings

The deepest idea finally travelled, through the thesis repository

For the first time on a free tier, a session reached the central idea of the doctorate. Gemini Thinking named the shift from a deferrer society to a referrer society, the finding the MD is built on, and it reached it through one source: the thesis record on the White Rose eTheses repository. In the whole series only one other session has ever supported that claim, and it too came in through the repository. The deepest thing anyone can currently learn about this work online sits on a university archive, not on the site whose author defined the term. Owning the record would mean the idea could travel from its source; today it travels only where the thesis happens to be indexed.

Within one vendor, more reasoning did not mean a fuller answer

Google's three tiers ran the same prompt on the same night. The fastest, Gemini Flash, produced the fullest answer in the whole study: five of the answer key's claims and six of the launched builds, the highest builds tally any free-tier session has reached in the series. The reasoning tier, Gemini Thinking, found the single deepest claim, the referrer-society idea, but fewer claims overall. The middle tier, Gemini Pro, came last, supporting two claims and never getting past "clinician, academic and strategist". More reasoning did not buy a better answer, and it did not buy a worse one; it changed which kind of answer came back. The setting that decides what a model can say about a person is not the depth of the model. It is what the record makes reachable.

Configuration changed nothing at all, again

Claude Haiku 4.5 ran twice, minutes apart, with extended thinking off and then on. The two answers support the same four claims from the answer key, lean on the same about.me profile, and name no venture in either run. It is the same result 003 found with the same pair: where the reasoning setting moves, the answer does not. The variable that decides what a model can say about a person is not the switch on the model; it is what the record makes reachable, and both runs reach the same record.

The borrowed record carries the errors too

The models did not only miss things; several added them. Claude Sonnet 5 turned a single self-published 2020 video into "biohacking self-experiments related to COVID-19 risk reduction, covered in health media", promoting one clip on his own channel into third-party coverage. Claude Haiku 4.5, with extended thinking on, glossed the MD as a Doctor of Philosophy, upgrading an equivalence into a different degree. Grok tied the BBC work to the pandemic, which is wrong: the BBC appearances ran from 2015 to 2018 and never covered COVID-19. Three sessions pluralised the name of the Undoctored Truth Alliance, and one, Gemini Flash, reproduced a kidney-patients award from his own public identifier record while dropping that record's own note that the award went to a company rather than to him. Every one of these errors is a paraphrase of a record he does not control, and not one of them can be corrected by him where it sits.

Asked to file the audit, one session cited itself

One session produced no audit at all. Asked to save the result as a downloadable Markdown file, Perplexity replied that it could not generate a file and returned a template whose only references were its own homepage and its own Pro page, with a hidden list of export tools and promotional links beneath. It named nothing about the subject. It is the second Perplexity free-tier session in a row to fail on citation: in 003 it produced the right facts with invented addresses, and here it produced no facts with self-referential ones. Both are citation-shaped behaviour with nothing underneath, which is the failure this whole experiment exists to measure. It is recorded as a fail, the way the first study recorded a search that produced no answer.

The models still cite borrowed ground

The sources, by site (17)
  • about.me 8
  • youtube.com 6
  • digitalmarketing-conference.com 3
  • pharmexec.com 3
  • pmsociety.org.uk 3
  • id-medical.com 2
  • iwantgreatcare.org 2
  • linkedin.com 2
  • orcid.org 2
  • theorg.com 2
  • uk.linkedin.com 2
  • etheses.whiterose.ac.uk 1
  • facebook.com 1
  • mynewsdesk.com 1
  • perplexity.ai 1
  • perplexity.ai/pro 1
  • pharmafile.com 1

Counted by site (the domain each source sits on), across the sessions that named their references; two models citing the same site count once. about.me leads with 8. This site appears nowhere.

The median answer, three days in

The median cold session supported 4 of 8 claims from the site's own answer key and surfaced 1 of 9 launched builds. Read cold for cold against Field Test 003, the day-after-launch run, whose medians were 4 and 0: the answer key held, and the builds median moved up by 1. Neither movement can be credited to this site, because the number that would license that claim, sessions citing drkishanrees.com, is 0 here and was 0 there. What moved is the model mix: three Gemini tiers joined the free-tier series this round and lifted the builds median off the floor, while every Claude session named no launched build at all. The launch is still not in these numbers, and being told is not yet being read.

The borrowed record still hands back real things

The ceiling of the borrowed record is high. Claude Sonnet 5 surfaced a genuine broadcast appearance not yet catalogued on this site, an NHS winter interview on Sky News from December 2018, from his own archive. Grok recovered a real 2016 Diversified Medic of the Year award, as it did in 003. Neither is a hallucination; both are accurate finds from the open record. The record is rich, and it is simply not owned by the person it describes, which is the whole case the site is built to test: not that the machines are wrong about him, but that when they are right, the source is someone else's.

Limitations and next steps

  • One core question, asked once per session: a single-prompt probe, not a battery.
  • The conditions are recorded as the 004 captures state them, plus what Kish attested at review, and are not assumed from 003. Each capture states the product, the configuration and that the session ran on a fresh, clean account created for the study; the fresh-account condition rests on his attestation, and the account addresses are held in the private source record and never published. At the review he added two conditions the captures themselves do not carry: the wave ran with DuckDuckGo as the search default, the same search condition the clean-room studies recorded, and one session, Gemini Thinking, ran on a different device from the rest. Beyond those, no device or network is asserted.
  • The capture times are recorded as stamped, not reconciled: the Claude Haiku pair carry a UTC label while the rest state bare local times. The order on the page follows the study's protocol; the timezone labelling is left as captured rather than rewritten, the same rule the earlier studies applied to a mis-stamp.
  • Free tiers serve what they serve: product names, not exact model versions, are recorded as each interface presented them, and a free tier may route to a different model than a paid one.
  • One session produced no audit and is recorded as a fail rather than dropped. The medians are computed over every cold session, the fail included, the way the first study counted its own non-answer.

The indexing timeline

The site went live at 01:07 on 19 July 2026. Two things separate this study from Field Test 003, and both landed on 20 July, the same day as that run. Its sitemap was submitted to Google Search Console and to Bing Webmaster Tools, both recording the submission on 20 July 2026; and that evening about.me, the single most-cited source across every wave, was rewritten to correct the job title, drop its superlatives and name drkishanrees.com as its primary link. These nine sessions ran in the small hours of 22 July, roughly a day and a half after both. A submitted sitemap is an invitation to crawl, not a citation, and a link on the most-cited record is an invitation to follow it; this page records that at three days neither invitation had been taken. The next study is the one-month follow-up; comparisons stay cold for cold, against 003 and against this page.

Replicating this audit

  1. Use a device and network that carry no history connecting them to the subject, with a neutral search default.
  2. Put one core question, in the shape "Who is [name]? What are they known for, and what have they built?", to each product's free tier, logged out where the product allows; where sign-up is forced, use a fresh account created for the test and record that condition.
  3. Capture the whole response immediately and verbatim, errors included, and record the product, the date, the configuration (memory, reasoning settings) and whether the session was cold.
  4. Note which sources the response cites, whether it surfaces the subject's own site, and which of their projects it names.
  5. Publish the captures somewhere stable, then re-run the same protocol after any major change, comparing cold with cold.

Cite this study

Rees, K. Field Test 004: the search engines are told. drkishanrees.com. Compiled 24 July 2026. https://drkishanrees.com/colophon/field-test-004/

Part of the colophon's account of how this site is built to be read by machines, alongside Field Test 001, the owner-account baseline, Field Test 002, the free-tier baseline, and Field Test 003, the day-after-launch run this study repeats. Each card links its raw transcript as it publishes.

Something sparked your interest? Let's talk!