Heritage Unlocked - Flipbook - Page 41
Translation, contextualisation and truth
Tom Chatfield
‘...how heritage is
represented in
data, and verifiably
anchored to
history, truth,
provenance,
human context,
is not a technical
nicety. It is
fundamental ’
I want to offer three ways of thinking
about technology and information in the
context of heritage: translation,
contextualisation and truth. Let me take
each in turn.
Translation. I use this word
deliberately, in part as an antidote to
talking about ‘data’ as if it had nothing to
do with life, values, hopes, heritage.
Data is a generic, fungible term, and it
disguises the fact that the things it refers
to may be real, tangible, personal.
What does it mean to bridge between
the human, cultural world and the
world of increasingly powerful
machines? The word translation
acknowledges something essential: that
such bridging is imperfect and partial.
Consider fragile illuminated
manuscripts, photographed in ultrahigh definition and made available to
scholars who can zoom in on their
details without risking damage to the
originals. This is wonderful. Technology
has even allowed us to read the letters
on burned and buried scrolls, texts that
were utterly lost to human knowledge
until imaging techniques revealed what
no eye could see.
These are genuine goods, as is the
democratisation of access to history and
heritage. Yet in the physical world,
objects come textured with size, heft,
weight, smell, provenance. All of that
dematerialises when we translate them
into the realm of data. Translation is not
simply loss, but nor is it equivalence.
Contextualisation. One of the
defining features of digitisation’s power
is that it makes everything grist to its
mill. AI, in particular, is ravenously
hungry for content. It reads every book
ever written. Or rather, it doesn’t read
them: it ingests them, represents them.
Every piece of art, every photograph, is
drawn into the same process. This is
tremendously powerful. But it is also a
profound act of decontextualization. It
offers us a view from everywhere and
nowhere.
Compare this with the human context
of being in a museum. This is a space we
occupy as citizens rather than
consumers; where we are invited to
watch, learn, respond emotionally to
heritage, the past, art, beauty. A
museum is a place that preserves and
provides context and particularity,
embodiment and uniqueness.
By contrast, think about what it
means for our primary mode of access to
a place to be via apps on the screen of a
mobile phone: navigation, user reviews,
updates competing for attention. What
does this do to our relationship with
places, time and history? The answer is
entwined with how far our use of
technology is thoughtful and
boundaried; and how those entrusted
with our heritage advocate for and
embody it within a digital age.
Truth. Systems like generative AI,
rather notoriously, don’t give you the
truth. They give you a plausible
response, a convincing reflection of the
data. You often get an answer that
pleases, or that persuades, but not
necessarily an answer that is right or
consistent. This is intrinsic to how such
systems work: they are probabilistic, not
factual; fluent, not knowing.
Ironically enough, this makes the
original, the physical, the known and
documented more powerful and more
valuable. The question of how heritage
is represented in data, and verifiably
anchored to history, truth, provenance,
human context, is not a technical nicety.
It is fundamental. The burden falls on
people to know what the system does
not (even cannot) know: to double down
on the truths of place, presence,
interrelationship.
This, for me, is the relationship that
matters most. What does it mean to
provide people with technological as
well as physical spaces and contexts
within which they can relate
authentically to their history, their
surroundings, their possibilities?
Opposite: The Founders’
Library at the Fitzwilliam
Museum, a space where
centuries of knowledge,
craft, and scholarship remain
accessible to all
Translation, contextualisation and truth
41
