There Is No Such Thing as AI Ethics
There is ethics.
It is older than the machines.
Older than the shareholders.
Older than the room
where the panel is being catered.
It was there before the badge lanyards.
Before the sponsored breakfast.
Before the woman at registration
found your name
and apologised for the spelling.
Aristotle took the stand.
He asked what sort of person
a person should become
before being handed power
and a procurement budget.
Kant adjusted the question.
He asked what we owed each other
when nobody was watching
except the system,
which was always watching
and had recently updated its privacy notice.
Mill ran the numbers
and asked who was doing the counting.
Confucius watched the conduct
of the clerk.
Weil measured the weight of affliction.
She did not mistake it for a variable.
The prophets asked about the poor,
as prophets inconveniently do.
Christ said,
do unto others
as you would have them do unto you,
which remains difficult
to monetise cleanly.
The dock had a queue.
The plough did not require
plough ethics.
The printing press did not require
printing press ethics.
The bomb did not invent
the ethics of killing.
The pill did not invent
the ethics of the body.
The internet did not invent
the lie.
Each new thing
was dragged back
to the old court.
Each new thing
stood under the same roof,
looked at the same judge,
and gave the usual evasive account
of itself.
AI is no different.
Who is told the truth.
Who is deceived.
Who is watched.
Who is sorted.
Who is priced.
Who is nudged.
Who is refused.
Who is displaced.
Who is enriched.
Who decides.
Who pays.
Who benefits.
Who is fired by a machine
whose reasons are kept
in a locked drawer
beneath a policy
nobody signed
but everyone obeyed.
Socrates would recognise
the smell of it.
So would Isaiah.
So would Du Bois.
Arendt knew the face.
Evil wore office shoes
and filed its paperwork
on time.
There is nothing new
in the inventory.
Only a new defendant
in the dock.
That is why
the new field arrived
with fresh carpet,
good lighting,
and no inherited memory.
No case law.
No old shame.
No settled judgement.
No grandfather
at the table
saying,
we tried this once
and it ended badly.
A new field
is a young thing.
Young things are easier
to sponsor.
Staffed by the people
being judged.
Funded by their foundations.
Reviewed by their fellows.
The journals founded
by the same hands.
The conferences named
by the same mouths.
The principles drafted
by the same desks.
The definitions approved
by the same rooms.
The minutes were taken. The sandwiches were adequate. The conflict of interest wore a clean shirt.
You cannot capture ethics.
It has too many ancestors.
Too many scriptures.
Too many dead people
still objecting.
Too many novels.
Too many mothers.
Too many workers
standing outside factories
with lunch wrapped in paper
and nothing abstract
about the rent.
But you can capture
AI ethics.
You can build it from scratch.
Lay the slab.
Name the centre.
Write the principles.
Invite the sponsors.
Publish the framework.
Approve the roadmap.
A field without witnesses
is easier to fence.
The most subversive sentence
in the room
has no dashboard.
It says:
We are not studying
AI ethics.
We are studying ethics.
AI is the current case.
Aristotle cannot be acquired
by Andreessen Horowitz.
The Sermon on the Mount
is not on a roadmap.
The Analects
do not have a Series B.
The principles were approved.
Nobody asked
who paid for the room.