Another AI Craze. Another Whopping Energy Bill. Burning Trust and Power with Video.

We've been so busy celebrating "democratised creativity" that we forgot to ask what we're actually creating—and what it's costing us.

This week's poster child for AI video generation landed with familiar fanfare: OpenAI's Sora, pitched as revolutionary creative democratisation, arrived like every other AI tool before it, with millions pasting in a prompt to generate "content" nobody asked for.

Within hours of Sora 2's 30 September 2025 launch, social media timelines filled with AI-generated clips. OpenAI showcased ‘Hollywood Blockbuster’ standard video where users could insert themselves into fantastical scenarios using the new "cameo" feature.

As soon as the announcement went mainstream, the race was on, not just to push the boundaries of possibility to create meaningful content, but to flood newsfeeds with synthetic imagery in the hope it captured that very rare online commodity: persistent viewer attention.

These visual tools are framed as democratising video production. In reality, they’re fuelling the latest wave of AI slop: unwatchable single-use junk clogging up timelines, inboxes and bandwidth, like discarded plastic washing up on the beach, unwanted and unwelcome. Powerful examples are lost in a sea of clones.

The real damage runs deeper than cultural pollution.

Recent 2025 studies into attention and energy usage suggest two big areas of concern: AI fatigue settling in from Ernst and Young and the IEEE warning data centres wasting much more electricity in part to feed resource hungry video content generators. 

The Environmental Bill Nobody's Talking About

A basic text query on ChatGPT consumes 0.34 watt-hours of electricity, according to OpenAI CEO Sam Altman. Generating a five-second AI video clip? Research by Hugging Face and MIT Technology Review found that requires 944 watt-hours—approximately 2,776 times more energy than a simple text prompt.

The International Energy Agency reported generative AI already consumes between 15 and 60 terawatt-hours annually. Data centres globally consumed 415 terawatt-hours in 2024. That's 1.5% of all electricity worldwide. It predicts usage could hit 1,000 terawatt-hours by 2026, with AI driving much of that explosive growth.

Every watt-hour spent on paddleboard backflips and dragon-riding cameos that sink into the algorithmic void is a watt-hour stolen from cancer research, traffic collision reduction, drug discovery. Work that actually saves lives.

Why is it a waste? Pure AI-generated content feels as unwelcome and valueless as a screenful of shouty animated banner ads from the late 90s.

Trust Is Burning Faster Than Electricity

The cultural cost mirrors the environmental one. As the 2025 Harvard Business Review research revealed, 40% of employees now report receiving “workslop”. That's AI-generated content so generic it actively wastes their time dealing with it. The average worker spends nearly two hours per month dealing with each instance, but the invisible tax goes beyond minutes lost.

According to Harvard's research, employees who receive AI-generated communications view their colleagues as less creative, less intelligent, and less trustworthy. Misusing AI to craft emails or content doesn't just create noise. It damages professional reputation, credibility and team rapport.

Outside the workplace, Ernst & Young Switzerland's research on AI content fatigue found that wider audiences instinctively recognise AI-generated material at first glance, since “having come to realise that most of it recycles known facts without adding much new insight, many ignore it without reading any further.”

This Isn't Harmless Fun Anymore

This isn't democratising creativity. It's businesses and creators dabbling with AI wasting resources to show off to audiences already tiring of fake content, filler fluff, and face-swapped fakery. While marketers scream "innovation," what is actually being built is distrust, deepfake fatigue, and data centres that devour electricity for swathes of vanity content nobody remembers.

It's infrastructure drain disguised as creativity. These throwaway clips burn real power, damage real trust, and add real friction for those businesses trying to do serious work augmented, not propped up or faked, by AI.

Before you paste another prompt into an AI video tool, ask yourself:

• Would I put my name to this in six months?
• Would I show it to a client to win more business?
• Is this worth the electricity it consumed to exist compared to simpler media?

Throwaway videos can't be recycled. They're the pinnacle of digital unsustainability.

AI slop doesn't just cost money. It costs trust. It costs credibility. It breaks down team rapport as poorly-trained AI fails to pull its weight. And increasingly, it's costing precious computational resources that could be devoted to groundbreaking research and development for medicines, and other pressing problems. Not deepfaked athletic somersaults on paddleboards.

We'd like your views:

• Should AI video tools display estimated energy consumption—just like fridges or flights?
• What’s the tipping point where "fun" AI content becomes environmental waste?
• How can businesses draw the line between valuable experimentation and slop?
• Do early adopters underestimated the reputational risk from low-quality AI outputs?
• When AI tools go mainstream, is human discernment optional—or more vital than ever?