A small field station on a hilltop, with a domed instrument tower and people at a table outside under big clouds, toned blue-grey under a faint grid
PL.01A field station•FIELD STATION
FIELD STATION· 54.9°N 2.1°W
fieldstation.xyz
A talk by FIELD STATION

Whose AI is it anyway?

Doug Belshaw

Tom Watson

Press H to see the keyboard shortcuts, S for the speaker notes

v0.34.2FIELD STATION LTD•
Way in

You don’t have to chat to use AI

A chat box, or work in the background Left: a chat window where you ask a question and the answer arrives at once as you wait. Right: a board of tasks, queued, running and done, that carry on when your laptop is closed. The chat box is "now"; background work is "soon" or "eventually". A CHAT BOX Answers at once. You wait for it. CHAT Summarise this report Ask anything → SPEED NOW WORK IN THE BACKGROUND Runs as you do something else. QUEUED Sort the inbox WAITING Draft newsletter WAITING RUNNING Read reports DONE Tag photos READY Check deadlines READY Laptop closed. The work carries on. SPEED SOON OR EVENTUALLY
Whose AI is it anyway?FIELD STATION LTD•
Three words worth knowingInside every AI tool

Model, harness, context

Context

What you give it

Your requestYour filesYour rules

Harness

The app around it

ChatGPTCopilotLM Studio

Model

The engine underneath

Closed
GPT-6 · OpenAIGemini · GoogleClaude Opus, Sonnet · Anthropic
Open
IBM GraniteLlamaMistral
Bearing helps you choose the model.RACK shapes the context.
Whose AI is it anyway?FIELD STATION LTD•
The evidence•Bearing

Only 26 of 354 Bearing tasks were chat

Bearing public dataset v2.1•354 tasks, 12 Apr to 2 Nov 2026•findbearing.org/data

354 tasks people brought to Bearing, one square each A grid of 354 squares. 26 blue squares are conversation. The other 328 are jobs: 171 solid squares need an answer now and 157 outlined squares can wait. CHAT JOBS ONE SQUARE, ONE TASK 26 328 230 of them recurring 171 need an answer now 157 can wait
Whose AI is it anyway?FIELD STATION LTD•
Thinking in layers•Strata

The same task can take a different route

One task, two core samples through the same three layers Three bands of rock: speed, where, scale. Two core samples are drilled through them. The usual default core reads now, Big Tech, individual. The just enough core reads eventually, your laptop, organisational. SPEED WHERE SCALE How soon? Who runs it? Who does it affect? USUAL DEFAULT Now Big Tech Individual JUST ENOUGH Eventually Your laptop Organisational
Whose AI is it anyway?FIELD STATION LTD•
Energy and carbon•Strata

Same energy, different carbon

GreenPT•when it runs

greenpt.com/carbon-aware-api
GreenPT's illustrative grid forecast over 24 hours: 238 grams of CO2e per kilowatt-hour at 16:00, 123 grams at the best start, 02:00

Waiting helps too. GreenPT’s own illustration: a 45-minute job due by 08:00 has about half the carbon if it starts at 02:00.

EcoLogits•where it runs•AI writing one email

4.17 watt-hours per email, in all three

Sweden
0.25 g CO2e
UK
1.01 g CO2e
Poland
2.55 g CO2e

Same email, same energy: ten times the carbon in Poland as in Sweden.

Estimates•CO2e: all greenhouse gases, counted as CO2•calculator.ecologits.ai, from the French non-profit Code Carbon•4 Nov 2026

Whose AI is it anyway?FIELD STATION LTD•
What it does•Bearingfindbearing.org

Describe the job. Bearing suggests a model.

  1. 1

    You describe the job in plain words.

  2. 2

    It weighs seven things:

    QualityCapabilityCostSpeedPrivacySustainabilityTransparency
  3. 3

    It ranks 81 models from 27 providers, and says why.

findbearing.org
Bearing's home page: one box, What do you want to use AI for?, with a job about case notes typed in, and a Find my model button

Free•no account needed•open source, MIT licence

Whose AI is it anyway?FIELD STATION LTD•
Choosing a model•Bearingfindbearing.org

Pick the model that fits the job

You tell Bearing

Bearing’s top pick

The trade-off

Case notes.
Must stay in-house.

GreenR (GPT-OSS)

OpenIn-house, on a big computer

Gain: privacy, openness, sustainability.

Lose: a little quality, and it costs a bit more.

Case notes,
on an ordinary laptop.

IBM Granite 4.0 Micro

OpenAbout 5 GB

Gain: cheapest, most open, runs on a laptop.

Lose: some quality.

Volunteer feedback.
Nothing about privacy.

OpenAI GPT-6 Luna Pro

Closed

Gain: the best quality.

Lose: openness, speed and sustainability.

Quality: how well Bearing expects the model to do this kind of job, from published tests and its own judgement•scores of 4 Nov 2026

Whose AI is it anyway?FIELD STATION LTD•
We’ve been here before

AI as a mainframe, or a network?

AI as a mainframe, or as a network Left: AI today, like a mainframe: one big machine in the middle with many screens wired to it, the way we all type into a few huge providers. Right: AI could be a network, many smaller machines linked to their neighbours and sharing the work. AI TODAY: A MAINFRAME A few huge providers. We all type into them. AI COULD BE A NETWORK Many small machines, sharing the work. MAINFRAME SERVER LAPTOP PHONE MAINFRAME TERMINAL
Whose AI is it anyway?FIELD STATION LTD•
Pooling devices•SWARMExperiment

Devices can pool to run one model

On its own, a device runs a small model. Together, twenty devices each run one slice of a model 100 times larger. Left: one device runs a small model by itself, like Qwen3 0.6B, about 0.4 GB. Right: the aim, not yet proven, is a model 100 times larger, like Llama 3 70B at about 40 GB, cut into 20 slices of 2 GB. Each of 20 devices in the room holds one slice. Your question goes into the first device, passes through every slice in turn, and the answer comes out of the last. ON ITS OWN Small A small model, on one device. Limited to what one device can hold. LIKE QWEN3 0.6B • ABOUT 0.4 GB TOGETHER • THE AIM: 100 TIMES LARGER, LIKE LLAMA 3 70B 20 SLICES, 2 GB EACH Slice 1Slice 2Slice 3Slice 4Slice 5Slice 6Slice 7Slice 8Slice 9Slice 10Slice 11Slice 12Slice 13Slice 14Slice 15Slice 16Slice 17Slice 18Slice 19Slice 20 ASK Each device holds one slice and passes the work on. A network, not a mainframe. ANSWER

swarm.fieldstation.xyz•built from SwarmLLM (now Pooled)•open source, MIT licence•tested with small models; 70B not yet proven

Whose AI is it anyway?FIELD STATION LTD•
Handing work on•RELAYEarly version

Close your laptop and the work carries on

RELAY is how we want work to move between devices when you leave.

Where the work goes

1

Your own devices

↓

2

A co-operative server

↓

3

Big Tech

Only if needed, and only after asking you.

Mock-up
RELAY
RELAY mock-up, handoff panel: Consultation summary. Laptop closed. Carried on at: co-op server. 26 of 40 responses. Timeline: started on this laptop at 16:42, lid closed at 17:04, picked up by a co-op server at 17:05, next your phone tells you when it is done.
Whose AI is it anyway?FIELD STATION LTD•
A court reporter in glasses writes in a notebook at a table as men stand and talk behind him, 1925
PL.05Court reporter, 1925•National Photo Co., LoC
Faster horses

AI meeting notes: a faster stenographer?

1925Court reporter, pen and notebook
2026AI note-taker, transcript and summary

Both give you one version of one meeting, without the wider context.

Neither links to the rest of the organisation

NOTES NO LINK DECISIONS WORKFLOWS HISTORY
Whose AI is it anyway?FIELD STATION LTD•
Your organisation’s rules•RACKtools.fieldstation.xyz/rack

Whose instructions does your AI tool follow?

You ask
“Summarise this month’s case notes.”
Usual default •Big Tech
Written byThe company that makes the tool.
Same for

Every organisation that uses it.

Just enough •your handbook, made with RACK

RACK by FIELD STATION (shared handbooks are early)•plain text files, a copy on each computer

Whose AI is it anyway?FIELD STATION LTD•
The same request, your rules•RACKIllustrative example

Same model. Different rules.

You ask “Summarise this month’s case notes”

Without a handbook•the vendor’s defaults

Sarah Jones (DOB 03/04/1981) was signposted to HB via the LA following a s.188 assessment. SW to f/u re UC sanction and DHP application.

With your handbook•made with RACK

  • Client A was helped to apply for housing benefit through the council.
  • Next: their support worker will follow up on a benefit sanction and a housing payment.
■ Required: no client names◧ Default: plain English□ Personal: bullet points

Your rules, not the vendor’s. Every tool that reads the handbook gets them.

Whose AI is it anyway?FIELD STATION LTD•
Who owns it?Via MozFest

Treat AI like a utility

“Co-operative models could work for smaller-scale infrastructure. Municipal ownership could work for community datacentres.”

Peter Wilson•4 Feb 2026•his own views

Co-op AI today

inference.coop

A member-governed co-op for AI. Open models, from $5 a month.

co/core

An experiment: members share spare computing power on their Macs to run open models for each other.

Water, electricity, and next, AI: things too important to leave entirely to the market WATER ELECTRICITY AI Next? Too important to leaveentirely to the market.

“AI Growth Zones: Gift Horse or Trojan Horse?”•peteski22.github.io•inference.coop•cocore.dev

Whose AI is it anyway?FIELD STATION LTD•
Seen from behind, a small figure reaches both hands across a large map spread on a table, toned blue-grey under a faint grid
PL.04Reaching across the map•FIELD STATION
The power of the default

Which of your tasks could wait until soon, or eventually?

Now Soon Eventually

Choices have consequences.

Try it: findbearing.org

Whose AI is it anyway?FIELD STATION LTD•
Whose AI is it anyway?Thank you

Which of your tasks could wait until soon, or eventually?

Questions?

FIELD STATION· 54.9°N 2.1°W

Doug Belshaw

Tom Watson

contact@fieldstation.xyz

fieldstation.xyz

findbearing.org

tools.fieldstation.xyz/rack

Whose AI is it anyway?FIELD STATION LTD•
Backup slidesClick one to open it
Whose AI is it anyway?FIELD STATION LTD•
ABearing’s public data

What Bearing records

It keeps the type of task and the choices made. It never keeps your own words, just labels Bearing writes about the task.

90%

recurring or batch

58%

were given a suggested series of steps

354 tasks, as of 4 Nov 2026•CC BY-NC 4.0

findbearing.org/data
Bearing public dataset page, including What we never collect
Whose AI is it anyway?FIELD STATION LTD•
BSoon and eventually

Tell it the deadline and it picks the time

GreenPT’s carbon-aware service picks a cleaner time to run, before your deadline.

They say it can cut a job’s carbon by up to 50%. Early access.

Find out more

greenpt.com/carbon-aware-api

greenpt.com/carbon-aware-api
GreenPT Carbon and Price Aware API page: Run AI when the grid runs green. One API that tells your agents when and where electricity is cleanest.
Whose AI is it anyway?FIELD STATION LTD•
CCode CarbonHeard about at MozFest

Measure your own AI’s carbon

codecarbon.io
Code Carbon home page: a French non-profit building free, open-source tools. Two cards: CodeCarbon for local compute emissions, EcoLogits for GenAI API impacts.

Computers you run

CodeCarbon measures the power, then applies your grid’s carbon.

AI services you call

EcoLogits estimates each request. Its calculator is on slide 6.

Free, open source•mostly for developers

codecarbon.io

Whose AI is it anyway?FIELD STATION LTD•
DMany models, working togetherSeen at MozFest

Many models, one answer

Panel and synthesiser

Several models answer. One combines them into a final answer.

Vote

Several models answer. The most common answer wins.

Chain

One model drafts, another critiques, a third revises.

Retake

Drafts are checked. If none passes, the models try again.

Route

Each question goes to the model that suits it best.

In development

Cascade

Start with a small model. Call a large one only if needed.

In developmentJust enough

Patterns from the ScreamingFace toolkit, backed by OpenMined, as shown by Dave Buckley at MozFest 2026•redrawn by FIELD STATION•screamingface.ai

Whose AI is it anyway?FIELD STATION LTD•
EMany models, working together•GitHubgithub.blog/changelog

Many models, working together: now built in

GitHub Copilot Fusion picks the models for each prompt. You pick the effort.

Three ways it combines them: one model on its own, a cheap draft checked by a judge, or a draft reviewed by a critic.

On all Copilot plans since 28 October 2026. You pay for each model it uses.

github.blog•changelog
GitHub's diagram, Three execution patterns in Copilot Fusion. Single: request, a primary model solves it with tools, answer. Cascade: a primary model makes a first pass, a judge model accepts it or escalates to a repair model that starts fresh. Critique: a primary model drafts, a read-only critic model approves it or asks for a revision, which the primary model makes.

Diagram: GitHub, “Route and orchestrate models with GitHub Copilot Fusion”, 28 Oct 2026•github.blog/changelog

Whose AI is it anyway?FIELD STATION LTD•
FMozilla.ai, Freehold, Otari

Run it yourself

freehold.earth
Freehold home page: Your data. Your rules. Your community.

Freehold: Peter Wilson’s project. Developer preview.

freehold.earth

docs.mozilla.ai
Mozilla.ai docs: any-llm introduction with tabs for other projects

Mozilla.ai: open tools such as any-llm and llamafile.

docs.mozilla.ai

Mozilla.ai’s Otari: one door to many models, with a spending cap for each team.

otari.ai

Whose AI is it anyway?FIELD STATION LTD•
GA local AI assistantMet at MozFest

Tiles

An assistant you own.

Runs open models on your own computer.

Chats sync between your devices, encrypted.

Backed by

Mozilla Foundation. An anchor project in its Human-Scale AI incubator, with $300,000, says Tiles.

tiles.run
Tiles home page: Own your AI. A local-first, collaborative AI assistant built with open models and decentralised protocols. Downloads for Apple Silicon and Linux.

Early alpha. Desktop app is a nightly build. Mac (Apple Silicon) and Linux only.

tiles.run

Whose AI is it anyway?FIELD STATION LTD•
HFrom TelepathSeen at MozFest

Television

AI doesn’t have to be a chat box.

A new kind of interface for your AI agent.

It builds dashboards on the fly, to suit the job.

At MozFest it was run entirely by voice.

Works with any model, including one on your own machine.

Open source, MIT licence•agent runs on Mac or Linux

television.run

television.run
Television mock-up: separate panels for a to-do list, a calendar and notes, with no chat window
Whose AI is it anyway?FIELD STATION LTD•
IPublic AI•AI Potluckalpha.aipotluck.org

Alpha’s X-ray shows the hidden choices

A free AI assistant from AI Potluck, an initiative by Current AI, a non-profit.

People bring the open models, datasets and tools. Its X-ray shows what happened behind each answer.

alpha.aipotluck.org

alpha.aipotluck.org
Alpha's X-ray panel for one answer: checked for safety, nothing flagged; read 6 web results; the model was Mistral Medium 3.5, made by Mistral AI in France; it ran on Mistral AI in France; the data is stored by Current AI in Alpha's own database, and guest conversations are deleted after 30 days

Alpha’s X-ray panel for one answer, 5 Nov 2026•AI Potluck, an initiative by Current AI•aipotluck.org

Whose AI is it anyway?FIELD STATION LTD•
JThe bigger pictureBased on the insight brief

Why does everyone default to Big AI?

Why does everyone default to Big AI? Two loops keep it that way; a third pushes back. Convenience loop: using Big AI by default makes Big Tech grow, which keeps it built into everyday tools and cheap for now, which means more use by default. Confidence loop: the more Big AI is used by default, the less hands-on experience people get of alternatives, so alternatives feel more daunting, so people use Big AI by default. Backlash loop: as Big Tech grows, the backlash grows and organisations try alternatives, which reduces default use and builds hands-on experience. Bearing works on alternatives feeling daunting. SWARM, RELAY and RACK give organisations alternatives to try. + + + − − + + + − + Built into everyday tools and cheap, for now Big Tech grows Use Big AI by default Alternatives feel daunting Hands-on experience Backlash grows Organisations try alternatives R CONVENIENCE R CONFIDENCE B BACKLASH BEARING SWARM • RELAY • RACK

Drawn by FIELD STATION from Mary Stevens’ insight brief•solid: more leads to more•dashed: more leads to less

Whose AI is it anyway?FIELD STATION LTD•
Swimming pool lanes seen from water level
PL.02A pool, not Exmouth•Artem Verbo, CC0, Wikimedia Commons
KLocal data centres

A data centre that heats a pool

Exmouth, 2023. Servers sit in oil. Their heat warms the pool water.

Lots of small sites, owned locally, could cope better than one giant one.

Source: AFP via Tech Xplore, June 2023

Whose AI is it anyway?FIELD STATION LTD•
LCredits and sources
  • PL.01A field station. FIELD STATION
  • PL.02A pool, not Exmouth. Artem Verbo, CC0, Wikimedia Commons
  • PL.04Reaching across the map. FIELD STATION
  • PL.05Court reporter, 1925. National Photo Co., public domain, Library of Congress
  • PL.05 shown in FIELD STATION duotone; PL.02 in its original colour.
  • ‘Just enough’ is Rachel Coldicutt’s framing.
  • Screenshots taken 4 and 5 Nov 2026 from findbearing.org, greenpt.com, freehold.earth, docs.mozilla.ai, tiles.run, television.run, codecarbon.io, alpha.aipotluck.org
  • Bearing data. Bearing Public Dataset v2.1, CC BY-NC 4.0
  • Copilot Fusion diagram by GitHub, from its changelog of 28 Oct 2026.
  • RELAY screens are a mock-up of FIELD STATION’s early version. Figures and the co-op server in them are illustrative.
  • Quote on the utility slide. Peter Wilson, “AI Growth Zones: Gift Horse or Trojan Horse?”, 4 Feb 2026. His own views.
  • Patterns on D redrawn from the ScreamingFace toolkit (screamingface.ai), as shown at MozFest 2026.
  • RACK by FIELD STATION. tools.fieldstation.xyz/rack
Whose AI is it anyway?FIELD STATION LTD•