The Reference Well Is Poisoned, But I Can Still Find What I Need.
Finding good visual references used to be the easy part.
You'd open Pinterest, drop in a few words, and spend twenty minutes going down a rabbit hole that eventually went somewhere. Images led to more images. Obscure corners of the internet surfaced things you hadn't seen before. You'd walk away with a moodboard that felt alive, specific, and genuinely useful for the work ahead.
That pipeline is broken now. Not completely, not for everyone. But enough that it's changed how I work, and I'd be surprised if it hasn't changed how you work too, even if you haven't named it yet.
What Happened to Pinterest?
Pinterest used to reward specificity. The more niche your search, the better it got. That's what made it useful for creative research. You weren't looking for popular. You were looking for precise.
The algorithm shifted. Sponsored content moved up. Trending content crowded out the specific.
And then the AI-generated images arrived…
They look plausible enough to land in results but have no source, no context, no actual designer or photographer behind them. They’re things that don't exist, presented as inspiration for the things you're trying to make real.
Pinterest added a filter to exclude AI-generated content. I use it. It helps. It doesn't fully solve the problem.
The deeper issue is that even the non-AI results have gotten noisier. The same images circulate widely enough that they stop being useful as reference. You're not finding something new. You're looking at what everyone else has already found.
Google Has the Same Problem
Search for a visual concept on Google Images and you'll get one of two things: a flood of unrelated results that technically contain your keywords, or an extremely crowded version of the most common interpretation of your idea.
Neither is useful.
What you actually need in the research phase is something that gets the gears turning. Something that isn't the obvious answer. Something that makes you think, or react, or push in a direction you hadn't considered. Google Images has gotten worse at producing that. The SEO-optimized, stock-heavy, AI-assisted results that rank well aren't the type of results that inspire good creative work.
The algorithm is optimized for clicks, not for creative usefulness. Those are very different things, and the gap between them shows up most clearly when you're trying to do real research.
- Pinterest has gone all-in on AI, flooding feeds with generated content and using AI moderation that's flagging real human work as synthetic. Users say it's destroying the platform's usefulness as a research tool. (Source: TechCrunch, 2025)
- An estimated 34 million AI images are generated daily, with roughly 70% of social media images now touching AI tools in some form. The volume has made it genuinely harder to find original reference that traces back to a real source. (Source: AutoFaceless, 2025)
- Creative ideation depends on exposure to diverse, unexpected stimuli. Research shows that when people engage with a wider range of visual experiences, their capacity for flexible and original idea generation increases. (Source: Creativity Research Journal, 2023)
Where I Go Now
I haven't stopped using Pinterest or Google. I've just stopped expecting them to do what they used to do.
Dribbble surfaces real work made by real designers. It skews toward UI and digital, but the craft is there, and it's a good place to see how other professionals are solving visual problems. I'm not looking to copy anything. I'm looking for a spark, a color relationship, a typographic choice, a layout decision that I can react to.
Stock Image Sites are underrated for inspiration. Not for the images themselves, but for the range. When you're searching a concept and you need to see twenty different visual interpretations of it quickly, stock sites deliver that in a way that's useful for early-stage research. I'm not licensing anything. I'm using the breadth of results to understand the territory.
Physical Sources still work. Books, printed materials, things that exist in three dimensions. The internet didn't make those obsolete. It just made people forget about them.
And then there's the move that I don't see many people talking about: the thesaurus.
A Thesaurus Is the Best Research Tool Nobody Uses
When the obvious keyword returns the obvious results, the problem is the keyword.
Before I search for anything visually, I'll open a thesaurus and find five or six different ways to describe what I'm actually after. Not synonyms exactly, but adjacent words. Words that approach the concept from a different angle. Words that might surface a different visual territory entirely.
If I'm searching for something that reads as "strong" and getting every generic result for that idea, I might try "resolute" instead. Or "uncompromising." Or "load-bearing." The image results will be completely different. Some of them will be useless. Some of them will be exactly what I needed, and I wouldn't have found them any other way.
This isn't a workaround. It's a research skill. It's the same thinking that goes into writing a good brief or naming a brand. Precision with language unlocks specificity in results. The creatives I know who are best at visual research are also unusually good with words. That's not a coincidence.
The Moodboard Disclaimer
I started adding a disclaimer to client moodboards. Not a long one. Just an acknowledgment that some reference images may be AI-generated, that I've filtered where I can, and that the moodboard is communicating a visual direction rather than a set of images to reproduce.
It's a small thing. But it changed the conversation.
Clients appreciate the honesty. It opens up a more useful discussion about what the moodboard is actually for, which is often misunderstood on both sides of the table. A moodboard isn't a mood recorder. It's a strategic communication tool. The images are doing the work of articulating something that's hard to say in words: a feeling, a tone, a level of refinement, a relationship between elements.
When you frame it that way, the source of the image matters less than what it's communicating. The disclaimer helps get there faster.
Disclaimer: Please note that the references used are not owned by FifthHouse and are sourced from Pinterest and other platforms.
What This Costs You
The research phase takes longer now. That's the honest answer.
Finding genuinely original visual references requires more effort than it did five years ago. You have to go further, search smarter, use more tools, and be more deliberate about what you're looking for before you start looking for it.
For experienced creatives, that's manageable. You've built instincts over years of looking at a lot of work, and those instincts help you evaluate what you find quickly. You know when something is interesting and when it's just familiar dressed up as interesting.
For less experienced creatives, or for clients who are doing their own visual research, the broken pipeline is a real problem. They don't always have the filters to recognize AI slop, or the language skills to search around the algorithm, or the reference pool to know when something is generic. They find what the algorithm surfaces and assume that's the territory.
That's part of why creative expertise matters in the research phase, not just the execution phase. The work of finding genuinely useful reference is real work. It requires judgment. And right now, the tools aren't making it easier.
Quick Answers
Is Pinterest still worth using for creative research?
Yes, with the AI filter on and lower expectations for discovery. It's more useful for confirming a direction than finding a new one.
What's the thesaurus trick?
Before searching visually, find adjacent or unexpected words that describe the concept differently. The image results will be different, and some of them will be more useful than anything the obvious keyword returns.
Why do moodboards include AI-generated images?
Because the filters aren't perfect and the volume of AI content on major platforms is high enough that it's hard to exclude entirely. A disclaimer sets expectations and shifts the conversation toward what the moodboard is actually communicating.
Does this mean the research phase takes longer?
Honestly, yes. Finding original visual reference requires more deliberate effort than it used to. That's a real cost, and it's worth accounting for in how you scope creative work.
The Takeaway
The visual research pipeline isn't what it was. Algorithms optimized for engagement, AI-generated images flooding major platforms, and SEO sameness crowding out the specific and the unexpected have made the early stages of creative work harder than they need to be.
That doesn't mean the work gets worse. It means it requires more intentionality. Knowing where to look, how to search, when to go analog, and how to use language to outmaneuver a broken algorithm are all real skills. They're part of what experienced creatives bring to the process.
The reference well isn't empty. It's just harder to find the good water now. And knowing how to find it, and being honest with clients when the conditions aren't ideal, is part of what good creative work actually looks like right now.
