How to give constructive feedback on AI-generated creative assets

So, your creative team has taken the plunge into using generative AI and a terrifying thought has suddenly gripped you —  are we actually creating good work, or do stakeholders, executives, etc. just think we are because it’s “AI magic”? 

It’s a legitimate question, because when your company churns out sub-par or average AI-generated content, you run the risk of not only demotivating your creative team, but also damaging your brand and becoming a laughing stock to customers.

And we’re not just talking about flagging extra fingers or weeding out copy hallucinations. The goal is to maintain the same level of quality you expect from human-created assets. 

But how do you do that? It’s simple — with feedback and revisions.

Below, we’ll get into how to leverage AI to your advantage without sacrificing creative quality, including how to give constructive feedback on these assets.

What we'll cover

Why you should use generative AI in the first place

“Generative AI has the potential to change the world in ways that we can’t even imagine. It has the power to create new ideas, products, and services that will make our lives easier, more productive, and more creative. [...] The future of generative AI is bright, and I’m excited to see what it will bring.” 

~Bill Gates

Gates may not know how to dance, but he gets tech trends. Roughly 59% of media agencies are now using AI-powered tools for some form of content creation (written or visual). AI is here to stay, and if you’re not fully exploring its potential for your creative workflows, you should. Here are a few reasons why it makes sense to use AI tools within creative projects.

Do more, faster

There was a time when people needed to sketch out things with their hands, and creative work had to go through a rough draft of some kind. Not anymore. AI can streamline the development of creative assets and assist with roles like graphic design by quickly creating variations of existing content, saving time and resources. 

Say you’re working on a fashion ad and can’t decide whether the model’s hair color looks better in black, blonde, or red. With AI tools, you can trial all three instantly. Even the most luddite, old-school designer can appreciate capabilities like that.

This type of efficiency is desperately needed: Around 92% of organizations are struggling to find the skills and talent they need to support their projects, which means more work falls on their current team members. AI can be a life preserver for your creatives who are drowning in tasks.

Cost savings

You can save a lot of cash by automating parts of the creative process using AI. Resources get freed up that can be allocated to higher-impact tasks, which means you can run leaner teams. At Ziflow for instance, we have creatives who use AI tools to assist with editing tasks for images, such as removing backgrounds — tasks that used to take them hours to finish manually.

Companies are already seeing significant savings by going this route. Klarna, for example, says it saved 37% on marketing costs by using AI in its processes. Creative agencies of all sizes can also see savings by reducing their workloads and serving more clients.

Inspiration and ideation

Creative blocks happen, and it can take a while for inspiration to hit. Generative AI design and content development can instantly give creatives new ideas to explore and allow them to easily experiment with different styles and formats.

Take a company trying to create a new logo for a new business or brand pivot. AI can generate several new logos for creative teams to look at and get ideas from. You can also take some of the generated AI output and use that to create new AI prompts to refine ideas until you find an idea to build on.

How do we put this lightly?… Sometimes AI seems to be under the influence. And it’s not just one substance, either — it’s like AI took all of them. This leads to tons of issues, like:

• Completely disregarding brand guidelines, giving customers an inauthentic (or worse, inaccurate) experience with your brand.
• Outputting incorrect information (often called “hallucinations”) that, at a minimum, damages trust in your brand, and at worst can land you in legal hot water.
• Not following prompt directions and generating content that doesn’t align with a project’s objectives.
• Creating generic content that provides customers with very little value, which could lead to a lackluster brand experience.

This means you’ll need strong review and feedback systems in place — lead by human creatives — to mitigate those challenges and produce high-quality deliverables. 

How to ensure your AI-generated content doesn’t make you look silly

If you know the weak spots of AI content, you can properly evaluate AI-generated content. Be on the lookout for:

• Accuracy and factuality: AI hallucinates more than roadie at a Grateful Dead concert. It gets information like dates, people, and events wrong. Review the facts on any asset created with AI to ensure everything is correct.
• Repetition and clichéd language: AI tends to use general terms, which often leads to repetitive and common phrases. These issues can reduce an asset’s quality, so look for places to remove repetition and use more diverse phrases. We recommend playing around with AI prompts and reading the outputs. You’ll start to notice the language patterns after a few tries.
• Content quality and creativity: AI isn’t really “creative” — it relies on the text and image data it’s fed to create new content. As a result, AI-generated content tends to be generic and usually looks like stuff that is already out there. So you’ll need to examine the text and visual assets generated by AI algorithms to see if it’s original and aligns with your creative standards and project goals.
• Alignment with brand guidelines: AI doesn’t always do a great job of adhering to all the guidelines you give it. Check the tone, style, colors, imagery, and messaging of AI output to verify that it aligns with your brand. Using AI doesn’t mean you can ignore brand compliance!
• Similarity to other work: AI models learn from large datasets — some of which may be copyrighted work. The copyright problem is getting so much attention there are now many legal cases about AI output and the data AI companies use to train AI models. Check your AI output against similar content to avoid any plagiarism concerns.

How to build a foolproof feedback process for AI-generated content

Now that you understand where AI tends to make a mess of things, you can use that knowledge to build an effective creative feedback process that focuses on those weaknesses.

As you follow the steps below to develop your AI feedback process, remember that the human element is the key to taking AI content from “meh” to something truly special.

1. Cultivate a transparent and truthful approach to AI content

AI is still new, so many creatives may be sheepish about or outright against using it in their workflows. This can lead to internal conflict and less efficient operations. So before you get started, explore potential use cases and decide if it makes sense to add AI to those processes.

If you decide to use AI in your organization, take a transparent and truthful approach. Don’t be afraid to talk about the possible applications of AI generation with your team and develop a plan that outlines when and where your team can use it. For example, maybe everyone agrees that using AI for idea generation is fine, but that using tools like DALL-E to create full-blown designs is not cool.

Be clear and make sure everyone understands that AI is a tool — not a crutch. 

2. Establish clear evaluation criteria

If you decide to use AI to create content, it’s time to develop evaluation criteria to review the AI’s output. Once you have a checklist of evaluation criteria, your team can build a process to review AI output. You can use the ideas below as a starting point:

• Check the content for brand compliance. Does it meet your client’s industry regulations? Does it use messaging, vocabulary, and product positioning consistent with their brand? AI-generated creative assets are notorious for missing the mark here.
• Fact- and accuracy-check using authoritative sources. You wouldn’t want your creative ad to say “Est. 1957” when your client has been in business since 1901!
• Consider your target audience. Will the content appeal to them? If the target audience is primarily Gen X professionals, you may not be well-served by AI content laden with language and visuals geared toward Gen Z.
• Verify usability for the target medium. Make sure any written content accounts for SEO optimization; visual content for social media should be engaging, etc.

3. Assemble a cross-functional review team with diverse expertise

Depending on your or your client's industry, you may need multiple people across different departments to review AI content and ensure it meets brand, marketing, and compliance standards. A diverse (see: human) team like this will bring a wide range of viewpoints to the process and help you optimize your AI content to its fullest potential.

This cross-functional team might include:

• Creative director: Ensures the assets align with the brand’s vision and quality standards.
• Subject matter experts: Fact-checks content for accuracy.
• Legal experts: Checks content for any issues, such as copyright infringement or unverified product claims, that may put the brand in legal jeopardy.
• Marketing specialists: Assesses whether the asset targets the intended audience and encourages the desired response.

4. Implement a structured feedback workflow

Once you have a reliable team together, it’s time to build an AI workflow that will help streamline the review process for AI assets. A typical workflow may include:

1. AI content generation: Creative teams generate the first draft of AI content and prepare it for review.
2. First round of review and feedback collection: Creative teams conduct the first round of reviews to ensure the AI content is high quality and meets the brand’s expectations. Using a centralized design feedback tool for feedback gathering can simplify the process and make it easier for all parties to review and leave comments on specific parts of an asset, streamlining the later revision process.
3. Cross-functional team review and feedback collection: This stage involves team members from other departments who verify that the asset doesn’t have any factual, legal, or quality issues.
4. Revise assets: The creative team now revises the assets based on the feedback. As with human-created deliverables, you may have to repeat the feedback and iteration steps several times.
5. Delivery and feedback collection: The finalized asset is delivered to the client. At this point, they’ll either approve it or request revisions depending on the scope of your engagement.

5. Facilitate feedback with a collaborative proofing platform

Using a centralized collaborative proofing platform like Ziflow to manage and track feedback for AI-generated content helps teams streamline the process, avoid missed steps, and work better together.

With a collaborative proofing platform, you can:

• Collect feedback in one place so creatives don’t need to go back and forth between their colleagues and clients to find information (no more searching through email chains and documents).
• Collaborate in real time without switching between several applications.
• Track changes, view revisions, and compare versions.
• Simplify approval process workflows to hand off assets to clients more quickly.
• Use automation for time-consuming tasks (like creating review assignments and sending reminders).
• Use templates to get started quickly instead of building new projects from scratch every time.

These features allow cross-functional teams to work together seamlessly, complete assets more quickly, and meet performance metrics.

Facilitate feedback seamlessly with Ziflow

The use of AI-generated content has shaken up the creative industry — in good, bad, and strange ways. Creative teams can harness AI to increase productivity, save money, generate new ideas, and even review content, but it has to be done correctly. Otherwise, your company will look silly to customers and your creatives will have one foot out the door.

You need the right review and feedback system in place to ensure that your AI-generated assets are high-quality, accurate, and fit the intended audience and project goals — and that’s where Ziflow comes in.

Ziflow helps teams manage their review process and build streamlined proofing workflows. With Ziflow, you can automate your feedback and approval workflows, collaborate in real time, and manage all your feedback in one place.