AI is everywhere at the moment.
Images.
Writing.
Music.
Search engines.
Coding.
Customer service.
Social media.
Video games.
And, naturally, everybody has an opinion about it.
For some people, AI is the future and anyone objecting to it simply needs to accept what is coming.
For others, the moment AI has been involved in something, the phrase “AI slop” appears almost immediately.
I do not think either position is particularly useful.
Because the problem with talking about “AI” as one enormous thing is that it really is not one thing.
Using AI to generate thousands of disposable images is not the same as using machine learning to help developers find bugs.
Cloning somebody’s voice without their permission is not the same as using software to automate a repetitive task.
Generating the artwork for a game is not the same thing as creating an enemy that can adapt to how you play.
So rather than asking whether AI itself is good or bad, I think there are better questions.
What is it being used for?
Who agreed to it?
What did it replace?
Was anyone told?
And did it actually make the finished thing any better?
AI Slop Is Real
Let us start with the obvious one.
AI slop exists.
We have all seen it.
Images pumped out with very little thought.
Articles that somehow use hundreds of words to say absolutely nothing.
Music generated in enormous quantities.
Videos with strange movements, odd voices and hands that appear to have been assembled during somebody’s lunch break.
There is a legitimate criticism there.
Generative AI makes it possible to produce huge amounts of content very quickly, and not everyone using it is particularly interested in whether that content is good.
Sometimes the goal seems to be simply:
Make more stuff.
Then make more stuff after that.
If somebody wants to call low-effort, mass-produced AI content “slop,” I understand why.
The problem comes when AI slop stops describing the quality of something and starts describing anything where AI was involved.
Those are not the same thing.
Something can be lazy.
Something can be badly made.
Something can be inaccurate.
Something can be disposable.
But if the only criticism is:
“AI touched this, therefore it is worthless.”
then we have stopped judging the work and started judging the label attached to it.
Most People Are Probably Somewhere in the Middle
The internet often makes these arguments look very simple.
You either love AI or hate it.
In reality, a lot of people seem to be doing something much less dramatic.
They are using it while also being worried about it.
That makes sense.
In June 2026, the Office for National Statistics found that 43% of adults in Great Britain thought AI brought roughly equal benefits and risks. Another 38% thought the risks outweighed the benefits, while 36% still expected AI to benefit them personally.
That feels considerably closer to reality than the idea that everyone has neatly chosen a side.
People can see obvious benefits.
AI can help search through information.
It can assist with repetitive work.
It can help with coding.
It can help organise things.
It can improve accessibility.
It can sometimes save a ridiculous amount of time.
At the same time, people are worried about jobs.
They are worried about misinformation.
They are worried about whether artists, writers and performers are being copied.
They are worried about companies replacing people because a machine is cheaper.
They are worried about whether they can trust what they are looking at anymore.
Those concerns do not cancel each other out.
You can find something useful while still questioning how it is being used.
That is probably where I sit.
Jobs Are Where This Gets Much Harder
One of the easiest ways to make the AI conversation pointless is to tell people worried about their jobs that they are simply afraid of progress.
That concern is not imaginary.
Companies will absolutely look at AI and ask whether it can reduce costs.
Sometimes that means helping people work more efficiently.
Sometimes it means deciding they need fewer people.
Those are very different outcomes for the person whose job is involved.
The International Labour Organization estimates that around one in four workers worldwide are in occupations with some degree of exposure to generative AI, but it also argues that most affected jobs are more likely to be transformed than simply made redundant because human input is still required.
That sounds considerably more believable to me than either extreme.
I do not think the realistic future is that AI suddenly replaces everybody and we all go home.
But I also do not think it is sensible to pretend nothing will change.
Jobs will change.
Some roles will shrink.
Some will disappear.
Others will be created.
And plenty will probably involve people working alongside AI whether they particularly wanted to or not.
The question I keep coming back to is fairly simple:
Is AI helping someone do their job, or is it mainly being used as an excuse to remove them from it?
That is not always easy to answer.
But it is a much more useful question than whether AI is inherently good or evil.
Creative Work Makes It Personal
This gets even more complicated when creativity is involved.
Art.
Writing.
Music.
Voice acting.
Design.
These are not just tasks on a spreadsheet.
People build careers and identities around them.
So when a company says it can generate a voice, image or piece of writing instead, the reaction is understandably stronger.
There is also a difference between AI helping a creator and AI imitating one.
If an artist uses a tool to speed up part of their workflow, that is one thing.
If a company trains a system on somebody’s work and uses it to avoid hiring them, that is another.
If an actor agrees to have their voice reproduced for a particular purpose, that is one conversation.
If somebody copies their voice without asking, that is a very different one.
That is why I think consent matters so much.
Consent Changes the Conversation
Voice actors have been one of the clearest examples of this.
The concern is not simply that technology can recreate a voice.
The concern is whether the person whose voice is being recreated actually agreed to it.
That distinction became important enough to form part of the 2025 SAG-AFTRA video game agreement, which introduced consent and disclosure requirements around AI digital replicas.
And it should matter.
If someone’s voice, likeness or performance is being digitally recreated, they should know about it.
They should have a say.
They should know what it will be used for.
That does not magically solve every ethical question, but it changes the conversation considerably.
Cyberpunk 2077 gave us an interesting example after the death of Polish actor Miłogost Reczek, who voiced Viktor Vektor.
For new material, another actor performed the dialogue and technology was then used to reproduce Reczek’s voice, with approval from Reczek’s family.
Some people will still be uncomfortable with that.
Fair enough.
But it is clearly different from secretly cloning an actor so you do not have to employ them.
Context matters.
Consent matters.
“They used AI” does not tell me enough.
Tell People What You Are Doing
Transparency matters for the same reason.
If AI-generated content is going into something people are buying, I think people should generally know.
Steam’s current content survey makes a useful distinction between AI-powered development tools used for efficiency and generative content that actually ships with a game.
It also separates content generated during development from content generated live while someone is playing.
That feels sensible.
I do not particularly need a developer to announce every piece of software used during production.
But if substantial artwork, dialogue, voices or other content were generated, that is relevant information.
Especially if the company knows some customers care about it.
Hiding AI use also creates another problem.
If people discover it later, the conversation stops being only about AI.
Now it is about trust.
And once people think you were trying to hide something, getting that trust back becomes considerably harder.
Not Every AI Tool Is Replacing an Artist
This is also where some of the discussion around gaming becomes too broad.
That includes things such as data handling, animation processing and other technical workflow improvements rather than simply generating finished game assets.
That is not the same as saying:
“We generated Resident Evil’s characters and fired everybody.”
It is still AI.
But it is solving a different problem.
There are plenty of parts of game development that are repetitive, technical or simply time-consuming.
If software can make those jobs easier while developers still make the creative decisions, I struggle to see why that should automatically be treated the same way as replacing the creative work itself.
The better question is:
Did AI replace the creative decision, or did it help someone get through the boring part faster?
Gaming Shows Why “AI” Is Such a Badly Defined Argument
Video games make this whole discussion especially messy because games have been using things called AI for decades.
Enemy behaviour.
Pathfinding.
Bots.
Companions.
Dynamic difficulty.
Procedural systems.
We just did not spend every conversation arguing about whether a Goomba was coming for someone’s job.
Generative AI has changed what people mean when they say “AI,” but even now there are several completely different technologies being bundled together.
A game might use AI for:
- testing
- animation
- generated dialogue
- artwork
- localisation
- smarter enemies
- adaptive difficulty
- NPC conversations
- development tools
So saying:
“This game uses AI.”
does not actually tell me very much.
What kind?
Where?
Why?
That is the interesting part.
Sometimes “AI Slop” Is Completely Fair
Then you get cases where the criticism is much easier to understand.
In October 2026, Jagex apologised after an externally produced RuneScape trailer included apparent generative-AI material.
One of the most noticeable problems was a character with six fingers.
Which is not ideal.
Especially when Jagex said the external partner had assured them generative AI would not be used.
That example contains most of the things people dislike about AI-generated content in one convenient package.
The quality problem was visible.
Oversight failed.
The company apparently did not intend for it to be there.
And the result was worse because of it.
If AI saves time but leaves you publishing a six-fingered character that nobody noticed, it seems reasonable to ask exactly what was improved.
That is a very different criticism from:
“AI exists, therefore bad.”
The problem there was the result.
AI Could Also Become Part of the Game
There is another side of gaming AI that I find much more interesting.
What happens when AI actually changes what we can do inside games?
KRAFTON’s PUBG Ally experiment uses AI to understand a player’s voice, interpret what is happening in the match, communicate and perform actions alongside them.
Whether that becomes genuinely useful remains to be seen.
It could be brilliant.
It could also spend half the match walking into a wall while confidently explaining that everything is under control.
That would at least be authentic multiplayer behaviour.
But the idea itself is interesting.
Imagine companions that actually respond to what you say.
Enemies that learn how you approach situations.
NPCs that react to events in ways that were not completely scripted beforehand.
Games already fake intelligence remarkably well.
Giving them systems capable of reacting in more flexible ways could create genuinely new experiences.
But there is still one very important test.
Is it actually better?
Technology being impressive does not automatically make the game fun.
If an AI character breaks immersion, talks nonsense or makes the experience worse, then it has failed.
It does not matter how clever the technology behind it was.
Gamers Do Not Agree Either
It is tempting to say gamers hate generative AI.
Some absolutely do.
There are gaming communities where the reaction to generative AI is overwhelmingly negative.
Particularly when it involves:
- artwork
- writing
- music
- voice acting
But broader surveys show a more mixed picture.
Plenty of players are neutral.
Plenty are unsure.
Some care deeply about how AI was used rather than whether it was used at all.
That makes sense.
Someone can dislike AI-generated concept art while being completely comfortable with AI helping test thousands of possible gameplay situations.
Those positions are not contradictory.
They are different uses of the technology.
We Are Now Seeing AI Where It Does Not Exist
There is another, slightly funnier consequence of all this.
People are now looking for AI everywhere.
A strange hand?
AI.
Odd lighting?
AI.
Something moves slightly strangely?
Definitely AI.
A trailer for Humankind 2 was accused of looking like “AI slop,” leading developer Amplitude to point out that it had been produced using real actors, practical sets and custom-made props.
And then there was my own considerably smaller contribution to the AI debate.
I posted a Mario video showing off some wall jumping.
Someone asked whether it was AI-generated.
It was not.
I can confirm that I was present when I recorded myself doing the wall jumps.
For one brief moment I apparently looked suspiciously competent.
So I will take that as a compliment.
But it does highlight something.
We have reached a point where AI is so visible in the conversation that people sometimes find it even when nobody used it.
That is probably not healthy either.
If everything slightly unusual starts being dismissed as AI, genuine human work gets caught in the backlash.
There Is an Environmental Cost Too
AI also needs physical infrastructure.
Data centres.
Electricity.
Cooling.
Specialised hardware.
None of that disappears because the technology feels virtual.
As AI use grows, so does the amount of computing power required to support it.
The International Energy Agency’s work on energy and AI projects substantial growth in global data-centre electricity demand this decade, with AI one of the major drivers.
That deserves scrutiny.
At the same time, I think this is another area where arguments quickly become exaggerated.
AI is not consuming all of the world’s electricity.
But it also does not have zero environmental cost.
Both things can be true.
A technology being used at enormous scale should be expected to justify the resources it consumes.
And, personally, it becomes much harder to defend those resources when the end result is thousands of pieces of disposable spam nobody particularly wanted.
Copyright Is Still a Mess
Then there is copyright.
Which is where everybody suddenly becomes a lawyer.
Including people who are very definitely not lawyers.
There are legitimate questions around:
- training data
- licensing
- ownership
- imitation
- human authorship
And many of those questions are still being worked through by courts, governments and copyright authorities.
The US Copyright Office has produced separate reports covering digital replicas, the copyrightability of AI-assisted work and generative-AI training, which gives some idea of how many different questions are hiding underneath the word “copyright” alone.
I do not think pretending the answers are already obvious helps anyone.
If AI systems are trained using existing creative work, it is reasonable for creators to ask how their work was used.
It is also reasonable to ask what rights somebody has over something produced with AI assistance.
Those arguments will continue for a long time.
Probably long after the rest of us are tired of hearing about them.
“AI Is Just a Tool” Is Not Quite Enough
One phrase that gets used regularly in defence of AI is:
“It is just a tool.”
There is truth in that.
But I think it is also a little too easy.
A tool can still be used badly.
And AI comes with questions that many traditional tools do not.
Where did its training material come from?
Is it imitating somebody?
Did that person agree?
Is it replacing someone?
How much energy is being used?
Can people tell whether the result is real?
Who is responsible when it gets something wrong?
So yes.
AI is a tool.
But saying that should probably be the beginning of the discussion, not the end.
The interesting question is what we are doing with it.
So How Do I Judge It?
I do not think I need one universal opinion on AI.
I can judge different uses differently.
For me, there are a few questions that matter.
What is the purpose?
Is AI solving an actual problem, or is it there because somebody wanted to say they were using AI?
Was there consent?
If somebody’s voice, likeness or work is involved, did they agree?
Was it transparent?
Are people being told honestly where generative AI was used?
What did it replace?
Is it helping someone work, or mainly being used to avoid employing them?
Where did the material come from?
Particularly when creative work is involved.
Did it improve the result?
Because if it made the game, image, article or whatever else worse, then all the technological cleverness in the world does not really matter.
And did somebody actually check it?
Because if your revolutionary AI workflow ends with six fingers making it into the final trailer, perhaps one human looking at the thing before publishing it would still be useful.
Judge the Use, Not Just the Label
Sometimes AI deserves criticism.
Sometimes AI slop is exactly what something is.
Sometimes companies use it badly.
Sometimes it damages trust.
Sometimes it cuts creative people out of work they should have been part of.
Sometimes it is used without proper consent.
Those things matter.
But they do not mean every use of AI is automatically the same.
A development tool is not the same as replacing an artist.
A dynamic NPC is not the same as cloning an actor’s voice.
Helping somebody work faster is not automatically the same as removing them.
And using AI somewhere in a creative process does not automatically mean there was no human creativity involved.
So when I hear that something used AI, that is not really enough information for me anymore.
I want to know what it did.
Who agreed to it.
What it replaced.
Whether anyone tried to hide it.
And whether it actually made the finished thing better.
Because AI is not a quality rating.
And neither is AI slop.
Sometimes the technology is the problem.
Sometimes the way it was used is the problem.
Sometimes it might actually be useful.
Judge what the AI did.
Not simply the fact that it was there.