Beyond AI Slop

How Brands Stay Credible as Content Gets Messy

TL;DR

We have seen this before

DSLR video put cinematic-looking footage into the hands of millions. Smartphones put a high-quality camera into the hands of billions. Both times, people declared the professional was finished. Both times, they were wrong because judgment, craft, and trust did not come bundled with the hardware.

AI is the same wave, but it moves faster and cuts deeper. It does not just lower the barrier to entry. It collapses the friction that used to slow bad decisions down. You used to need time, skill, or budget to publish at scale. Now you need none of those things. Which means if your judgment is poor, or your process is loose, you can be wrong at scale before anyone in your organisation notices.

The difference now is not whether you can produce content. The question is whether it is worth someone’s attention and whether it is safe to put your name on it.

Why it is moving faster than most teams expect

Two forces are colliding.

1. The unit cost of “good enough” is collapsing
Standford Univeristy‘s AI Index reports that the cost of querying a model at roughly GPT 3.5 level performance dropped from $20 per million tokens in November 2022 to $0.07 per million tokens by October 2024, a reduction of more than 280 fold in around 18 months.

2. The culture has already named the downside.
Merriam-Webster named “slop” its 2025 Word of the Year, defining it as low quality digital content produced in quantity by means of artificial intelligence.
That is not a niche internet joke anymore. It is a mainstream signal that audiences are developing a sharp allergy to low trust media.

Platforms are also leaning into it. Youtube’s CEO Neal Mohan, said that in December alone, more than one million channels used YouTube’s AI creation tools. The tide is strong, as Kapwing’s research found that 21-33% of YouTube’s feed may consist of AI slop or brainrot videos.

The question, then, is not whether to use AI. It is how to use it without quietly leaking the trust you have spent years building.

Outputs vs systems

AI makes outputs cheaper. That part is unavoidable. But outcomes still come from systems. If you are buying “a video”, the market will keep price pressuring that category, because tools keep getting faster and “good enough” keeps getting easier.

If you are buying a system that reliably produces outcomes, you are buying the things AI does not magically solve:

• Message discipline and story strategy
• A repeatable workflow that protects quality
• Governance, approvals, and version control
• Rights clarity, disclosure choices, and reputational safety
• A cadence that builds trust over time
• An evidence loop, what worked, what did not, and what to do next

Without that system, AI does not make you more effective. It just helps you publish faster, including faster mistakes. This is the line worth keeping: AI does not remove the need for judgment. It raises the cost of bad judgment.

How we approach AI – FoSho!

We use AI across our workflow, but we do not use it to replace jobs.

We use it to elevate the work, unlock creative ambition, and make previously unrealistic projects possible within real world time and budget constraints. We are also deliberate about context. Different organisations have different governance, risk profiles, and stakeholder realities. Sometimes AI plays a bigger role. Sometimes it barely touches the process. Either way, our quality bar stays the same.

Some of the questions we ask ourselves are:

1. Is the content true, and can we evidence it
2. Is it on brand, and approved by the right owner
3. Do we know what data touched what tools
4. Do we know what we own, and what we licensed
5. Who had final human oversight and approval on this?
6. Would we feel comfortable defending this publicly

One more thing that matters in practice: the tools change fast. What we used last quarter is not necessarily what we use today. So our approach has to stay insight led, tested, and documented, not trend led. Our job is to keep you ahead of the curve without sacrificing credibility, or accidentally becoming the next cautionary headline.

The impact is here

It is tempting to frame AI as a future jobs story. For most people and organisations, it is already a present-day task and process story.

The World Economic Forum Future of Jobs reporting expects significant skills disruption by 2030, and highlights that a large share of workers will need training and upskilling in the coming years.

The more useful question for leaders is: are we upgrading our people and processes fast enough to stay credible, productive, and competitive as the tools change?

We like to summarise it like this: If your job is primarily actioned in front of a screen, then AI will fundamentally change the way you work. It might happen slowly, but as some stage, it might happen very fast.

UK analysis from IPPR has warned that a meaningful share of tasks are already exposed to generative AI capabilities, and that exposure rises sharply if organisations integrate AI more deeply, which increases the risk of displacement without employer action and policy support.

This is why the “augment – not replace” stance only works if it is backed by training, guardrails, and good management.

Training is not a nice to have. Real world deployments show measurable time and productivity gains, especially for less experienced staff, but only when usage is guided and quality controlled.

One widely cited field study from the International Labour Organisation in customer support found productivity gains in the mid teens on average, with much larger gains for novices. A UK government update on AI use in the Civil Service reported average time savings of around 26 minutes per day in a trial, which adds up quickly across a year.

The pattern is consistent: AI can compress time, but without guardrails it can also compress trust.

What to look for in a content partner using AI

If you are evaluating agencies, partners or your own organisation, these questions could help cut through the buzzwords:

1. Where exactly does AI touch the work, and where does it not?
2. What organisational and/or client data is uploaded to which tools, and what is the retention stance?
3. How do you avoid low-quality output and content for content’s sake?
4. What is your disclosure approach, and how do you align it to our brand risk?
5. How do you manage rights, licences, and provenance in a mixed workflow?
6. How do you measure success, and what does “worked by AI” actually mean?

A thoughtful, experienced person should be able to answer all of these clearly, without defensiveness, and without promising miracles. If they cannot, they are not ready to carry your brand risk.