Dan Sodergren

I “Authentically Inspire” organisations about the future of work, AI and the Fifth Industrial Revolution | TEDx and keynote speaker, media guest and author | Trusted by Kelloggs, Sodexo and more...

I’m a professional speaker who inspires organizations and their leaders about the Future of Work, Artificial Intelligence, and the Fifth Industrial Revolution. 

As a TEDx speaker, BBC commentator, and media expert, I’ve spent decades in digital marketing, AI, and technology. Making complex ideas simple, practical, and inspiring. 

My work has featured on BBC World, Channel 4, ITV, and leading podcasts and radio shows.

I’ve founded and grown multiple startups — from YourFLOCK.co.uk (helping people be happier at work) to the AI Marketing Course and AI Teacher Course, empowering businesses and educators to thrive with AI. 

I’m also proud to champion diversity and inclusion in tech, working with organisations like the MPA and The Big IDEA.

At the heart of everything: I’m a dad, entrepreneur, and optimist about the future. I believe technology, used wisely, makes work more human, not less.

👉 If you’re looking for a keynote speaker who energises audiences, challenges thinking, and leaves leaders with actionable insights — let’s connect.

www.dansodergren.com  www.aimarketingcourse.co.uk  www.greatmarketingworks.co.uk

Co-Founder

Your FLOCK investigates an individual's value profile in your team. And quickly online collates your true #companyculture map.

To help managers and bosses  increase #workingfromhome happiness, retain key talent and increase #remoteworking team productivity.

As #companyculturematters more than ever.

Dan Södergren

Full-Stack Data Engineering & AI Lifecycle. Compact Data Science - No 1 Mathematical and Statistics Skills on Amazon (June)


Working at the intersection of systems and stories — designing the architectures that shape our digital future, while keeping sight of the values and narratives that make it worth building.

I operate across three connected domains:

- Foundational AI and machine learning research, exploring the structures, representations, and emergent dynamics of intelligent systems. Forthcoming book set with World Scientific.

- Computational simulation and policy research, modelling technological change, market power, and institutional futures. Forthcoming book with leading educational publisher.

- Practical automation and agentic workflows, turning advanced models into real-world tools that augment human capability. Forthcoming book with Packt.

Alongside this technical work, I maintain an active visual and conceptual art practice, approaching creativity as a complementary form of systems thinking — a way of making abstract ideas tangible and humane.

As an Advisory Board member of the World Innovation and Change Management Institute (WICMI), I contribute to shaping research agendas around transformation, governance, and responsible innovation. Most recently, we published an AI governance manifesto exploring the need for new institutional frameworks:
👉 https://www.linkedin.com/pulse/towards-digital-constitution-manifesto-ai-era-political-economy-aqzqf

My writing and research examine the deep relationship between political economy, digital platforms, and artificial intelligence — particularly the concentration of computational power, control over behavioural data, and the economics of algorithmic extraction.

For essays and analysis on AI, power, and platform capitalism, subscribe to my Substack, Alogopoly:

"Who really owns OpenAI's surplus?" 
https://substack.com/@alogopoly/note/p-180160178

“When openness becomes a weapon” 
https://substack.com/@alogopoly/note/c-176231540

“The next great knowledge hijacking” 
https://substack.com/@alogopoly/note/p-179141053

“Why Europe didn’t build Google” 
https://substack.com/@alogopoly/note/p-179543974

Through AICognitia’s mentoring programs, I help leaders apply AI thoughtfully — building strategies, workflows, and organisational literacy that combine technical depth with critical thinking.

Areas of Focus:

Machine Learning Research | Computational Simulation | AI Automation | Governance | Innovation | Digital Transformation | Futures Thinking | Visual Storytelling


Angel Salazar, PhD

A new paper from Stanford dissects the privacy policies of Amazon, Anthropic, Google, Meta, Microsoft, and OpenAI.

Its core finding is deceptively simple: all of them train on user chat data by default.

For years, debates around AI training data focused on web scraping but chatbots represent a structural escalation. Instead of scraping what we published, firms now ingest what we privately disclose.

Chat interfaces encourage disclosure. Unlike search queries — sparse, transactional — chats are dialogical. They unfold over time. They accumulate context. They invite reflection, vulnerability, experimentation. Users upload documents, health details, financial questions, drafts of legal disputes, intimate dilemmas.

The Stanford authors show that, across the board, these interactions are incorporated into model improvement pipelines unless users actively opt out. The default assumption is consent.

Several firms retain chat data indefinitely or for extended periods. Even when chats are deleted, their informational imprint may persist within trained weights.

Once conversational data becomes part of training corpora, it becomes infrastructural.

The paper also highlights something more subtle: product boundaries are dissolving. Data from search, email, social media, cloud documents, and AI chat interfaces increasingly converge.

This challenges core principles embedded in the California Consumer Privacy Act — data minimisation and purpose limitation. When every product is an AI training vector, purpose expands to encompass the entire platform ecosystem.

We are witnessing the consolidation of "cognitive surplus extraction": not just harvesting clicks or attention, but harvesting structured reasoning traces, memory preferences, and latent identity signals.

Another underexamined finding: enterprise customers are generally opted out of model training by default. Consumers are not.

The asymmetry is even sharper regarding minors. Some platforms allow teen users and may train on their data, raising questions about consent validity and long-term digital footprints.

The Legitimacy Question

Developers frame chat training as a social good: “improve the model for everyone.” And indeed, more data often improves model performance.

But legitimacy requires symmetry.

If individuals bear the privacy risk, they must have meaningful, informed, frictionless control. Opt-in regimes, shorter retention, robust filtering of sensitive data, and transparent training disclosures are not anti-innovation. They are conditions for trust.

If chatbots are to become cognitive infrastructure, then their data practices must be treated as infrastructural governance — not UX fine print.


#Alogopoly #DataPrivacy #SurveillanceCapitalism #AIgovernance #LLM #Chatbots #DigitalPower #PlatformEconomics #AIethics #PurposeLimitation #CCPA #AIpolicy #InformationAsymmetry #TechOligopoly #BehaviouralData #CognitiveExtraction #AIRegulation #DigitalRights

Author, AI governance scholar & artist exploring how digital architectures shape power and perception, while advancing foundational ML research, policy simulation, and agentic AI workflows for social impact.

taplio

The Battle

Dan Södergren

Angel Salazar, PhD

Followers

Posts

Engagements

Likes

Performance ScoreAngel Salazar, PhD's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Dan Södergren

Angel Salazar, PhD

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Angel Salazar, PhD's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreAngel Salazar, PhD's LinkedIn Performance