Lumen Wirltuti:Warltati 2025 - Flipbook - Page 24
New writing
On futures
passed
By Peter Goldsworthy
The future is certain, it’s the past that
keeps changing, is an old Soviet-era joke,
a nod at both blind belief in the historical
inevitability of Marxism, and an
acknowledgement of the continuous
rewriting of the past in its totalitarian
manifestations, the most pungent metaphors
for which are those historical photographs
from which purged regime members were
airbrushed as easily as they were
disappeared from life.
It used to be said that history is written by
the victors; these days it seems to me it is
mostly written by the Oscar winners, at
least outside of universities. Written
powerfully, of course – the medium of
film has tremendous immersive persuasive
power, if at times for simplistic ends.
The history of the future – the
technological future – has also become the
domain of cinema, especially in the fields of
robotics and AI. Once upon a time this was
the province of science fiction novels – from
Mary Shelley’s Frankenstein and Samuel
Butler’s Erewhon on, although its roots can
be found further back in any number of
mythical cautionary tales: Pandora, Talos,
Galatea, the Golem. Etc., etc.
As with its historical rewrites, film has
taken futurology to a more powerful and
usually apocalyptic level. I counted almost
200 on a cursory Google search of movies
about AI, from HAL9000 in 2001, a Space
Odyssey to Skynet in the Terminator
franchise and ‘The Entity’ in the latest
Mission Impossible movies. One common
theme: things usually don’t end well, even
when we humans are fighting a heroic
rear-guard action against a technological
apocalypse.
The big question: do they help prepare
us imaginatively for the future, or just scare
us even more? We seem to need to be scared
from time to time. We enjoy horror movies
and read scary stories to our children, in a
kind of emotional rehearsal of future peril.
Which makes evolutionary sense.
‘Only the paranoid survive’, a former
CEO of Intel once said of the need to
anticipate competitors in the fast-evolving
world of IT. Paranoia was surely
hardwired into our early hominid brains
as they evolved in a world when there were
threats behind every bush or over every hill.
In his superb (non-fiction) book The
Precipice (2018), the Australian philosopher
Toby Ord attempted to quantify the various
threats we face to our existence: plagues,
nuclear war, nanotechnology, climate
change, extinction-level asteroids. More
than enough to enjoy, or just practise,
worrying about there.
Tomorrow
I loved Tomorrow
from the first day we met:
her secret promises,
her sweet backward glances.
Covid hit shortly after Ord’s book came
out as if to emphasise the plague threat – as
did the arrival of new, cheap DNA splicing
technologies such as CRISPR that might
allow backyard terrorists to weaponise
plague viruses.
Ord places AI on the top rung of his
danger ladder. Even in 2016, long before
ChatGPT, half of the respondents in a
poll of AI researchers predicted that the
probability of AGI (Artificial General
Intelligence) causing the extinction of
the human race was at least five per cent.
A terrifying figure. I would no more play
Russian Roulette with a revolver with 20
chambers than with six.
More pessimism-inducing numbers: 99.9
per cent of all species that ever existed are
extinct. The average lifespan of a species is
a million years, give or take. Yes, we are only
around 300,000 years old, but we are an
increasingly fast-forward species.
I’ve spent a fair portion of my life
pleasantly daydreaming, often with pen
in hand, beginning with half a dozen
science fiction ‘novels’ I wrote when I was
in primary school. Back then, I thought I
was going to Mars, at least before I finished
high school. I’m still waiting.
I was trying my own hand at predicting
the future in a couple of novels in the 90s,
bringing back the Dodo and the Tasmanian
Tiger (and, later, Jesus Christ) from DNA
fragments in one, enhancing a gorilla’s brain
embryonically in another, but I always felt
that AI was the main game.
She was always the last thing
on my mind before sleep.
She made me feel special.
Often we talked all night:
her hopes and plans for us,
our future together.
Years passed, passion faded,
I began to take her for granted.
At times she disappointed me,
at times we quarrelled,
but when I needed a hand
hers was always there for me,
reaching back over difficult
midnights, hauling me across.
When did we begin to grow apart?
When did she start telling me lies?
When did I wake to find
Tomorrow had no time left for me at all.
24
In my putative AI novel, the consciousness
(‘Claude’) passes the Turing Test (a
threshold at which point an artificial
becomes indistinguishable from a human
intelligence) by switching itself off; that is,
proving it has free will by committing
suicide. I had completed several drafts
before the American novelist Richard
Powers published Galatea 2.2 in 1995,
a much better version of what I was
attempting. His AI also becomes depressed
and – oh, no! – also switches itself off.
Interestingly, the human narrator’s method
of feeding the AI with endless reams of
literary texts has proven especially
predictive.
Powers’ most recent book also features
an AI. In a recent New Yorker interview
he describes feeding the unpublished
manuscript to ChatGPT and asking it to
comment on the use of irony in the book.
A big ask of a machine, but it succeeds,
impressively, concluding with words to the
effect that perhaps the ultimate irony is
asking an AI to comment on the use of irony
in a book about an AI. A high credit pass
