Researcher / InnovatorParis, France

MichaelAnslow

I build new technological experiences, embedded in research and informed by a wider societal vision.

Between systems and stories

A practice grounded in engineering, expanded through research and philosophy.

I turn early ideas into working technological experiences. I take early-stage AI ideas from vague opportunity through research framing, data and model development, interactive prototyping, evaluation and communication. My work is deliberately practical: research questions are explored inside systems that people can use, test and respond to. Across natural language processing and multimodal AI, I have built systems for knowledge representation, document exploration, public discourse analysis, creative writing and text–audio modelling. This includes an AI writing assistant used by professional musicians and Funiki, which extends generative technology into dynamic light-and-sound experiences for physical spaces. I also lead collaborations, communicate complex research and create spaces where technical and human questions can meet. I co-founded an interdisciplinary AI and philosophy community, and contribute technical support and research to discussions about the potential role of AI in education.

01 / Inspiration
Vague intuitions
Observation · Possibility mapping · Initial hypotheses
02 / Conceptualisation
Articulated concepts embedded in research
Literature review · Concept framing · Research questions
03 / Systematisation
Systematised understanding and requirements
Principles · Requirements · Data, model and evaluation design
04 / Implementation
Concrete implementations
Prototyping · Integration · Testing and iteration

Experience, methods & tools

The technical and research methods I use to make ideas concrete.

Experience, Methods & toolsTechnical range, organised by purpose

Language AI

Representing, generating and retrieving meaning

Transformers & language modelling
Semantic embeddings
Information retrieval

Multimodal ML

Learning across representations and media

Contrastive learning
Audio–text modelling
Generative modelling

Research & experimentation

Turning uncertain ideas into testable directions

Problem framing & literature synthesis
Rapid prototyping
Evaluation & experiment tracking

Software prototyping

Building interactive research systems

Python
JavaScript & TypeScript
Django, Flask, Dash & Godot

Data & retrieval

Structuring, searching and connecting data

PostgreSQL & MongoDB
Solr & FAISS
Neo4j & graph data

Delivery & reproducibility

Making experiments reproducible and systems deployable

Docker & Docker Compose
CI/CD
AWS & Weights & Biases

Selected work

Projects that move between artificial intelligence, creative practice and collective inquiry.

01Research creation · Sony CSLRole: Originator

Funiki

Object-based immersive lighting

02Evidence-based art · Research creationRole: Creative technologist

Feeling Belonging

Research-creation driven installation

Research creation · Related to Funiki

Feeling Belonging

Finding a connection to nature through a handmade cloud

Feeling Belonging is a distinct evidence-based art installation growing from the same questions as Funiki: how can light and sound alter our felt relationship with a space? Here, that question becomes an exploration of belonging to nature through an artificial weather object.

A handmade cloud anchors a simulated environment streaming the changing light of a day-and-night cycle. Mapped LEDs, illumination and reactive audio rise and fall together through a slow breath cycle, giving the installation a living rhythm without trying to imitate nature literally.

In collaboration with Zéphyr LorneCognitive scientist & experience designer

The completed illuminated cloud suspended above a seat in the exhibition installation — Exhibition installation / 01The completed installation, Feeling Belonging: a cloud, a seat and a space to breathe.

How the experience breathes

One continuous system, from virtual sky to physical light.

01 / Object: A handmade cloud with individually mapped LEDs.
02 / Environment: A simulation streams a full day-and-night light cycle.
03 / Rhythm: Reactive audio and illumination follow a shared breath.

Prototype journal / 01—05

Making the cloud was part of understanding it.

We moved between small material studies, lighting tests, software simulation and full-scale fabrication before assembling the exhibition installation.

Cloud armatures, electronics and fabrication materials on a workshop bench — 01Armatures and mapped lighting take shape on the workbench.

A hand testing the translucent mesh and fibre structure inside the cloud — 02Building a translucent outer layer that could hold and diffuse light.

A small illuminated cloud prototype beside computers running a three-dimensional simulation — 03Connecting the physical prototype to its simulated environment.

A person filling a large illuminated handmade cloud with fibre — 04Giving the full-scale cloud its volume by hand.

Two collaborators carrying an early cloud prototype through a garden — 05Moving an early cloud body between collaborative tests.

03Community · ParisRole: Co-founder & organiser

AI–Phi

Paris-based community on AI & Philosophy

AI × Philosophy × Community

A room for the questions technology cannot answer alone.

AI–Phi is a Paris-based community of researchers and professionals united by an interest in the intersection of artificial intelligence and philosophy. We track the ideas shaping a rapidly changing world through seminars, participatory conversations and informal community activity.

I am a co-founder and organiser of AI–Phi. I help shape the programme, facilitate discussion and create a space where technical, philosophical and societal perspectives can genuinely meet.

40: Sessions reached by June 2026
03: Seminars, causeries and other sessions
Paris: A local interdisciplinary community
My role: Co-founder & organiser

01 / Seminar series

Bring expertise into the room

Invited researchers and practitioners present a position, then open it to sustained interdisciplinary discussion.

02 / Causeries

Think together, not at an audience

Member-led conversations inspired by the Parisian salon use shared questions, readings and short contributions to examine a topic.

03 / Community

Continue beyond the session

News updates, informal exchanges, drinks and dinners create the trust needed for ideas to travel between disciplines.

Causerie output / 15 January 2026

Hopes, fears, trust, understanding and regulation

A collective response to an open discussion about the gap between engineering uncertainty and public expectations of AI. The resulting contributions examine autonomy, persuasion, ethics, trust and the limits of regulating emergent systems.

Includes Michael's contribution on the limits of pragmatic evaluations of intelligence.Read the collective output

Edmund Ronald's live sketch from an AI–Phi session — Creative community / Live sketchEdmund Ronald's live sketch made during the session with Dominique Lestel.

Members of the AI–Phi community gathered for session 26 — Community / Session 26Snapshot of one of our sessions at Sony CSL Paris in the Quartier Latin.

Selected conversations

Ideas we have put on the table.

Across forty sessions, the programme has moved between minds, machines, values, language, agency, regulation and the material foundations of computation.

04Creative assistance · Sony CSLRole: Originator

Poiesis Studio

Pre-ChatGPT masked writing assistant

05Side interest · Ongoing inquiryRole: Co-author & tech volunteer

AI & Education

Research and volunteering on AI & Education

Other work

Previous projects, collaborations, research threads, wider practice.

Multimodal AI · Sony CSL Music Team

Text & Audio Project

Role: Text lead

Led the text work for an internal Music Team collaboration with data engineer Amaury Delort, exploring multimodal deep learning across language and audio, including NLP pipelines, language modelling, MongoDB and a Dash interface for querying and curating datasets.

Technical practice: Python & NLP pipelinesLanguage modellingMongoDB & Dash
Ways of working: Technical leadershipCross-team collaborationData curation

Data exploration · Research system

Charting Data Landscapes

Role: Originator

A research platform for navigating complex information through semantic representations, retrieval pipelines, data processing and visual interfaces.

Technical practice: DashSOLRInformation retrieval
Ways of working: Project conceptualisationPartner collaborationInterface framing

Public-interest NLP · Rapid research

Sony CSL COVID Language Team

Role: Project lead

Computational-linguistics research examining public discourse across languages and national contexts during the first wave of the pandemic.

Technical practice: NLP pipelinesData analysisVisualisation
Ways of working: Project leadershipRapid researchCross-language collaboration

Platform engineering · EU research

Penelope

Role: Software engineer

First prototype of an online platform, PENELOPE, for composing OpenAPI-described services into analysis pipelines to facilitate the development of distributed components across several institutions. Created for the European Union project ODYCCEUS.

Technical practice: DjangoOpenAPIRESTFull Stack
Ways of working: Research collaborationTechnical documentation

Research

Selected publications across music AI, natural language processing, knowledge representation and data analysis.

The through-line is an interest in how complex representations become useful, navigable experiences.

Full profile on Scholar

Portfolio ends / Writing beginsAn outlet for exploring and crystallising ideas

Beyond the work

Enter theblog

A place to think in public through essays and creative outputs. Some pieces become speculative targets for research—making ideas concrete enough to question, test and develop.

Open the writing site

Selected writing05 essays

Dec 2025

Materialising Myths

Exploring future societal roles of AI

Nov 2025

The Light and Dark of Awe

Part II — Aw-ful

Oct 2025

The Light and Dark of Awe

Part I — Awe-some

Jul 2025

A Vision of Writing Assistance Before ChatGPT

Part II — The Signal in the Machine

Jul 2025

A Vision of Writing Assistance Before ChatGPT

Part I — Poiesis Studio

Browse the complete blog

Have an unusual idea?Let's make it tangible.

MichaelAnslow

Language AI

Multimodal ML

Research & experimentation

Software prototyping

Data & retrieval

Delivery & reproducibility

Funiki

Feeling Belonging

AI–Phi

Poiesis Studio

AI & Education

Text & Audio Project

Charting Data Landscapes

Sony CSL COVID Language Team

Penelope

Enter theblog

Materialising Myths

The Light and Dark of Awe

The Light and Dark of Awe

A Vision of Writing Assistance Before ChatGPT

A Vision of Writing Assistance Before ChatGPT

Let's connect.