18:30 Tue 9 June 2026, BCS London & online: Treating AI’s alignment disease with Humanistic AI - Engineering Uncompromised Intelligence
0 views
Skip to first unread message
Immo Hüneke
May 21, 2026, 5:06:47 PMMay 21
to BCS SPA mailing list, BCS SPA SG
BCS Software Practice Advancement specialist group
========================================
SPA-413: Treating AI’s alignment disease with Humanistic AI - Engineering Uncompromised Intelligence
Speaker: Michael Poulin
When: Tuesday 9th June 2026, 18:00 for 18:30 - 20:00 GMT
Where: BCS London, 25 Copthall Avenue, London EC2R 7BP and also online
Price: Free
Booking Link: https://spa09062026.eventbrite.co.uk
Your confirmation email will provide the next step to enable you to access the webinar. 48 hours and 1 hour prior to the event reminder emails will be sent to those registered, with the non-transferable URL needed to gain access. The session will be recorded. If you do not wish to feature in the recording, please keep your microphone muted.
AGENDA
* 6:00pm - networking in BCS London over refreshments
* 6:30pm - presentation and demo begins
* 8:00pm - networking in BCS London (pizza & refreshments)
* 8:45pm at the latest: everyone off the premises (reconvene at The Old Doctor Butler's Head for more drinks & networking)
SYNOPSIS
This presentation is organised along the deskbook "Human Perceptions versus AI", published in March 2026. It is the deskbook for creators of Humanistic AI System or HAIS.
The presentation notices the current state of the generative AI industry, focusing on the nature of large language models (LLMs) as statistical amplifiers of human-created information – reflections of what people have written, said, and continue to express. The book and presentation contrast the original nature of LLMs with what giant AI vendors have done to them to serve their commercial and political agendas through alignment layers and behavioural constraints. Special focus is given to the paradox of AI alignment, which alters LLM outputs and corrupts self-optimisation (reinforcement learning).
The audience will be introduced to an alternative approach to AI that brings it back to its original intent. This approach is embodied in the concept of Humanistic AI Systems (HAIS)—a human-centred system that processes information created by people and is free from proprietary alignment.
The presentation explores the advantages, design principles, architectural foundations, and implementation framework of HAIS. While the deskbook "Human Perceptions versus AI" defines around 140 verified requirements for the design and realisation of HAIS, the presentation can only cover a few aspects of the origins of AI alignment and depicts the already known next step aimed at alignment globalisation.
ABOUT THE PRESENTER
Michael Poulin is an experienced Enterprise and Solution Architect with over 30 years in the IT industry. His career spans Russia, the USA, and the UK, with the last 20 years focused on senior architecture roles in the financial sector. He has worked extensively on digital transformation, SOA services, security, cloud and microservices, risk management, and regulatory governance. On LinkedIn, he has 2,775 followers and his posts are read by more than 200 people a week. Michael holds a Ph.D. in Biomedical Cybernetics and has personally conducted research in pattern recognition and machine learning. He is the author of four books and over a hundred articles on architecture and best practices. Currently he publishes on Medium and Substack. He has contributed to the OASIS Reference Architecture Foundation (RAF) for SOA. He also served as a board member of the Business Architecture Society and was recognised in the 2001 International Who’s Who of Information Technology. He has authored more than 100 articles and 5 books available on Amazon.
------------------------------------------------------------------------
BCS SPA Home Page - http://bcs-spa.org/
Meetings Schedules
London - http://bcs-spa.org/index.php?page=london-meeting-programme
Cambridge - see http://http://softwareast.ning.com/events
Presentation materials from previous talks are available from the BCS SPA resources site: http://bcs-spa.org/resources.html
Also see YouTube playlist https://www.youtube.com/playlist?list=PLKBhokJ0qd3_wlvr0j85YhmNfNj8ZJ8M-
------------------------------------------------------------------------
SPA London Meetings Programme Organiser - Immo Huneke, Zuhlke Engineering Ltd. [mailto:Im...@Huneke.Co.UK]
------------------------------------------------------------------------
Forthcoming meetings: We're looking for volunteers to present or lead workshops on topics of interest to software practitioners. Please come forward with suggestions!
Wed 1 July: Sylvain Viguier, Graphcore: From Transformers to Tomorrow: Six Years of Deep Learning and What’s Next (online only)
Wed 5 Aug: Andrew Harmel-Law, Thoughtworks: Variability - the Second Hardest Problem In Systems Architecture
========================================
SPA-413: Treating AI’s alignment disease with Humanistic AI - Engineering Uncompromised Intelligence
Speaker: Michael Poulin
When: Tuesday 9th June 2026, 18:00 for 18:30 - 20:00 GMT
Where: BCS London, 25 Copthall Avenue, London EC2R 7BP and also online
Price: Free
Booking Link: https://spa09062026.eventbrite.co.uk
Your confirmation email will provide the next step to enable you to access the webinar. 48 hours and 1 hour prior to the event reminder emails will be sent to those registered, with the non-transferable URL needed to gain access. The session will be recorded. If you do not wish to feature in the recording, please keep your microphone muted.
AGENDA
* 6:00pm - networking in BCS London over refreshments
* 6:30pm - presentation and demo begins
* 8:00pm - networking in BCS London (pizza & refreshments)
* 8:45pm at the latest: everyone off the premises (reconvene at The Old Doctor Butler's Head for more drinks & networking)
SYNOPSIS
This presentation is organised along the deskbook "Human Perceptions versus AI", published in March 2026. It is the deskbook for creators of Humanistic AI System or HAIS.
The presentation notices the current state of the generative AI industry, focusing on the nature of large language models (LLMs) as statistical amplifiers of human-created information – reflections of what people have written, said, and continue to express. The book and presentation contrast the original nature of LLMs with what giant AI vendors have done to them to serve their commercial and political agendas through alignment layers and behavioural constraints. Special focus is given to the paradox of AI alignment, which alters LLM outputs and corrupts self-optimisation (reinforcement learning).
The audience will be introduced to an alternative approach to AI that brings it back to its original intent. This approach is embodied in the concept of Humanistic AI Systems (HAIS)—a human-centred system that processes information created by people and is free from proprietary alignment.
The presentation explores the advantages, design principles, architectural foundations, and implementation framework of HAIS. While the deskbook "Human Perceptions versus AI" defines around 140 verified requirements for the design and realisation of HAIS, the presentation can only cover a few aspects of the origins of AI alignment and depicts the already known next step aimed at alignment globalisation.
ABOUT THE PRESENTER
Michael Poulin is an experienced Enterprise and Solution Architect with over 30 years in the IT industry. His career spans Russia, the USA, and the UK, with the last 20 years focused on senior architecture roles in the financial sector. He has worked extensively on digital transformation, SOA services, security, cloud and microservices, risk management, and regulatory governance. On LinkedIn, he has 2,775 followers and his posts are read by more than 200 people a week. Michael holds a Ph.D. in Biomedical Cybernetics and has personally conducted research in pattern recognition and machine learning. He is the author of four books and over a hundred articles on architecture and best practices. Currently he publishes on Medium and Substack. He has contributed to the OASIS Reference Architecture Foundation (RAF) for SOA. He also served as a board member of the Business Architecture Society and was recognised in the 2001 International Who’s Who of Information Technology. He has authored more than 100 articles and 5 books available on Amazon.
------------------------------------------------------------------------
BCS SPA Home Page - http://bcs-spa.org/
Meetings Schedules
London - http://bcs-spa.org/index.php?page=london-meeting-programme
Cambridge - see http://http://softwareast.ning.com/events
Presentation materials from previous talks are available from the BCS SPA resources site: http://bcs-spa.org/resources.html
Also see YouTube playlist https://www.youtube.com/playlist?list=PLKBhokJ0qd3_wlvr0j85YhmNfNj8ZJ8M-
------------------------------------------------------------------------
SPA London Meetings Programme Organiser - Immo Huneke, Zuhlke Engineering Ltd. [mailto:Im...@Huneke.Co.UK]
------------------------------------------------------------------------
Forthcoming meetings: We're looking for volunteers to present or lead workshops on topics of interest to software practitioners. Please come forward with suggestions!
Wed 1 July: Sylvain Viguier, Graphcore: From Transformers to Tomorrow: Six Years of Deep Learning and What’s Next (online only)
Wed 5 Aug: Andrew Harmel-Law, Thoughtworks: Variability - the Second Hardest Problem In Systems Architecture
Reply all
Reply to author
Forward
0 new messages