PB February-March 2026
February-March 2026 57
By Gerard Cunningham
AI: five synonyms for slop
T
he post told a simple story. A
parent teased two nine-year-olds.
Telling them the family would
have dinosaur meat for dinner.
The kids reply: “That’s AI”. In
previous lives they might have said “pull the
other one”, “yeah, right” or perhaps “fake
news”.
When an industry builds a product so
shoddy that pre-teens use it as a synonym for
fake and dubious information, that industry is
in trouble.
The ‘AI industry’ has been in speculative
bubble airspace for some time now, as anyone
who follows Ed Zitron’s ‘Where’s Your Ed At?’
newsletter can attain.
When tweenies have more sense than the
venture capital burning billions chasing their
silicon dreams of immortality and basilisk
gods while burning fossil fuels at suicidal rates
to power a hallucination engine, something is
very wrong in the world.
Over one fifth (22%) of the electricity
generated in Ireland goes to power the 80+
data centres dotted around the M50 hosting
thousands of servers handling massive
amounts of cloud computing, storage and AI
demand.
EirGrid calculates the figure will rise to over
30% by the end of the decade, driven mostly
by the AI bubble.
From 2017–2023, data centres absorbed
half of all new renewables (and all new
wind) generation, preventing net fossil-
fuel reductions. Their electricity demand
exceeded their long-term-contract renewable
procurement, while many are turning to on-
site gas generation, increasing emissions and
creating a major blind spot in Ireland’s already
dismal climate and energy projections.
Electricity demand from data centres
increased by 412% between 2015 and
2023. By comparison, server farms in
other EU countries typically account
for 3% of electricity consumption.
Yet somehow, everyone from venture
capitalists to newsroom proprietors and
editors persists in thinking AI is the Next Big
Thing. The emperor has no clothes. Youngsters
can see it. Only adults are in denial. Or
perhaps, as Upton Sinclair once observed,
“it is dicult to get a man to understand
something, when his salary depends upon his
not understanding it”.
There is a fundamental contradiction at the
heart of the relationship between journalism
and the large language models (LLMs) of so-
called artificial intelligence.
The magic pixie, the demon in the machine,
produces statistically likely strings of words
with no regard to whether or not they are true
— in other words, slop.
They just have to sound plausible. It is, in a
very literal sense, a bullshit generator.
Newsrooms are — or for the most part aspire
to be — the opposite of bullshit (or whatever
you want to call it: AI would no doubt furnish
five instantaneous synonyms). Journalism
is about, in the words of Joe Friday, “Just the
facts, ma’am”.
Every statement in a news story should be
verified, or at least, verifiable. The currency of
a newsroom is reliability and trust.
The aim of good journalism must be
trustworthiness, built on a foundation of truth-
telling. But the lying pixie cares only about
truthiness – blobs of truth-shaped wordage
strung together based on statistical models
and enough unsustainable power to heat a
city.
It is possible that AI may some day be
capable of getting over the bullshit problem
but most academics believe the tendency to
confabulation is mathematically embedded
in the models and systems they have built. For
now, barring the discovery of new algorithms
to counter its inherent flaws, fabrication is a
feature, not a bug. And it infects every corner
of the internet, poisoning websites, search
engines, and social media networks. For
example, roughly one in five videos promoted
by YouTube’s AI-driven recommendation
algorithm is unusable “slop”.
Yet despite the self-evident contradictions,
it seems not a day goes by without a major
newsroom announcing a licensing deal with
AI.
Worse again, some use the cut-price
junk machine proudly. Gannett, the largest
newspaper publisher in the US, and Insider,
a major digital news outlet, cut editorial sta
in 2023–2024 while expanding AI-generated
or AI-assisted content in sports reporting,
earnings coverage, explainers, and routine
news summaries.
This is a grave error. The experiments
worked poorly: automated articles attracted
criticism for errors and low quality, damaging
credibility and forcing both outlets to scale
back or heavily supervise their AI use.
News outlets have long invested in building
trust with their readers, listeners and viewers.
More than ever, that trust is the new currency.
Advertisers have an incentive to link their
products with quality and trust. That’s why
they seek out endorsements from trusted
athletes, musicians or movies stars, after all.
No one selling a product where reputation
matters wants to be associated with slop.
Reliable news, prominently and proudly
declaring itself an AI-free zone, willing to tell
viewers to slow down, wait, and take a moment
until the facts are checked before publication,
becomes increasingly valuable in a zone
flooded at scale by the garbage machines. This
can build a loyal base willing to support and
pay for reliability.
This is not complicated. Even a nine-year-old
can see it.
Journalism needs trust not algorithm-based
truthiness and pervasive hallucination
From 2017–2023, data
centres increased
electricity consumption in
Ireland by 412% absorbing
all new wind generation
and preventing net fossil-
fuel reductions
MEDIA
Village_FebMarch26.indb 57Village_FebMarch26.indb 57 03/02/2026 08:2403/02/2026 08:24
