Mention AI to creative workers, and you might be greeted with skepticism or fear. The idea of using a generative AI model to write copy, draw a picture, or compose music is a challenge to the conventional thinking that only humans can bring artistry into the world. It can also feel threatening to people in creative industries and business functions.

But, for now, AI is not a threat to those who are open to learning how to use it. It can speed up the initial steps in the creative process, whether that’s a first draft of a marketing campaign or a sketch for a new product. It can also automate and speed up content personalization, distribution, and testing.

In fact, the prospect of producing better creative output more quickly is resonating with businesses across industries. In a recent survey of nearly 600 chief marketing managers by Everest Group, more than half of the respondents said that content creation was the most frequent use of AI. “Many of our clients are telling us [that] generative AI has allowed them to produce content more quickly,” says senior analyst Aakash Verma.

However, it’s essential to understand what AI can and can’t do. Creative staff must review and improve the AI-generated material, ensure its accuracy, and imbue it with the context and emotional spark needed to deliver results.

Additionally, as vendors roll out new AI capabilities, and as businesses learn how to use (and not use) AI, any best practices for AI use could be out of date by the end of 2025. Creative individuals and their managers must implement an “explore and experiment” mindset and adapt their way of thinking as the technology and our understanding of it change.

Here’s what businesses need to know about incorporating AI into the creative process. This includes giving staff the time and freedom to learn how to use it, implementing guardrails that minimize security and other risks, and changing workflows to maximize the benefits.

AI’s place in the creative process

First off, it’s important to understand that by itself AI cannot create original content. Anything it generates is derived from existing content, whether that’s a collection of Renaissance paintings, this year’s pop hits, or your company’s brand guide. By digesting all this existing content, however, it can produce a first draft far more quickly than a human could.

This can reduce an onerous, time-consuming part of the creative process. Advertising teams, for instance, can spend days or even weeks working through the first ideas of a creative brief for a new campaign. AI-generated drafts—even if they’re far from usable—allow creative staff to move on much faster to the real magic of the creative process: injecting the material with human artistry, punching up the emotional impact, or aligning it with the brand message or target audience.

“Normally it would take our product designers 30 or 45 minutes to come up with good copy for an element or feature within the product,” says Bernard Meyer, Senior Director of Communications and Creative at marketing automation vendor Omnisend. When that’s multiplied by three to five times per week, he says, “AI provides significant time savings.”

Omnisend recently used AI to create a video script for a PR research project, prompting the tool to work from a press release and other relevant information, including audio and visual cues. “We then proofread and made a few changes to the script, all within two hours,” he says. “Normally, this process would take two to three days.”

AI can also help creative professionals translate their visual, text, or musical ideas into a tangible format in seconds rather than days when using conventional digital design tools. This allows them to share their ideas more quickly and effectively, get feedback, and fine-tune the output.

“The technology can also use information about a customer’s past actions to provide faster, more precise personalization than traditional rules-based approaches, and speed up the testing of content across distribution channels,” says Everest’s Verma.

What AI can’t do—but creative staff can

If businesses skip the human review process, it can expose them to the risk of lost credibility, brand damage, and even legal action if the content is amateurish, inaccurate, or contains stolen intellectual property.

“If you care about quality and brand, never expose your users to the outputs of AI without being transparent that it’s AI,” says Jake Moshenko, CEO and co-founder of authorization and access control vendor AuthZed. “Even if it’s right 99% of the time, the 1% who are having an absolutely abysmal experience are going to blame your company.”

AI can also inject errors. “On more than one occasion, I’ve asked AI how to accomplish a goal within our human resources information system, and it will generate a set of steps that look and sound entirely reasonable. But when I go to perform the actions, sometimes the pages, sections, and buttons it references don’t even exist,” says Moshenko.

Human oversight is also essential for the seemingly simple task of quickly summarizing existing articles or reports. An AI tool will “look for, say, four main ideas, and if there are only three main ideas in the paper, it will give you a fourth idea anyway,” says Anthony Miyazaki, a marketing professor at Florida International University. “Or if I ask for the five main ideas [and] there are only two, it will still give me five.”

How to optimize AI in the creative process

Every business has its own needs, risk appetite, budget, and skills regarding AI. However, there are four key issues to consider when using AI for creative functions.

1. Address the potential legal concerns regarding data ownership.

A hotly debated issue today is whether AI models can legally incorporate the work of writers, musicians, and visual artists into AI-generated materials. Some argue that AI is stealing and reusing intellectual property without permission, proper sourcing, or remuneration. Others say AI follows a historical precedent of artists and authors building on the work of others.

While lawmakers, lawyers, and creators ponder the issue, some businesses are protecting themselves by developing creative datasets of content they own to avoid any intellectual property (IP) infringement issues.

In other cases, AI vendors are certifying their training data or providing some customization of their models for various use cases, which reduces the need for customers to retrain them with new data. Some vendors, such as Microsoft and Adobe, use Content Credentials, which provides information about the origin and history of AI-generated content. Adobe also compensates creators whose content was used to train its Adobe Firefly generative AI machine learning model. A number of startups are developing marketplaces or licensing models that will compensate creators for data used in training AI models.

Businesses should work with their technology, legal, risk management, and compliance offices to devise a governance approach for managing IP issues and other AI risks, such as inaccurate output. But these efforts can’t get in the way of slowing the creative staff’s experimentation and innovation efforts with AI. The AI compliance function should work in the same way as a road manager does for a musician: They let creative people do what they do best, without worrying about whether the amplifiers are loaded on to the truck or about how the food service will be set up.

2. Educate creative professionals and their managers on how to work with AI.

Businesses need to be up-front with their creative staff about the risks and benefits of AI. Managers should candidly tell their creative staff that they’ll need to work with AI to stay relevant, Miyazaki says, and offer training to help them adapt. While many people will learn by doing, they will need additional support.

For example, to get the most precise and useful output, they’ll need to learn how to refine the requests they make of AI. Rather than asking the model to “Write me three headlines,” for instance, they should be more specific: “Write me three headlines in the voice of a young, irreverent copywriter, make it short, and give me two versions, one for a young consumer, the other for a middle-aged tech buyer in England.”

Creative professionals also need training to look beyond the creative aspect of their work to its business objectives, such as click-throughs or sales. “Many designers come at their work from the artistic side,” says Miyazaki. “If you’re going to do your work for, say, an advertising firm, your job is to figure out whether, when, and how your output changes audience behavior.”

Such creative types also need to be taught what AI cannot do, which is where they will need to spend their time. Beyond adding an emotional or original “punch” to content or tailoring it to a specific need, this could mean providing quality control and serving as a “creative engineer” who sits between the creative and the technical groups to help both sides make the most of AI.

Employees will also train themselves if you give them the opportunity, adds Moshenko. One method used at SAP is an AI “sandbox,” a safe engineering environment that allows your staff to experiment with different generative AI tools to see examples of their output before sharing it with the world.

3. Fortify your data management and security processes.

AI requires accurate, complete, and bias-free data, whether for training an off-the-shelf or in-house model. Regular audits, validation tools and processes, and integration of siloed data are all needed. Data governance and tracking data lineage can also help trace and explain why an AI tool suggested the outputs that it did.

Data security is another essential element. One of the biggest threats is accidental leakage of sensitive information, says Moshenko. Such leaks can occur when retraining an existing model with a smaller, more targeted dataset to make it better suited for specific tasks and when using external data for training with processes such as retrieval-augmented generation (RAG).

RAG is a particular threat, says Moshenko, because it imports data without preserving the access controls. Businesses need to consider scalable ways to govern access control, such as an authorization engine that replicates the permissions for the original data to ensure the user who supplies the prompt has access authority before that data makes it to the AI model.

4. Work closely with business, technology, and creative managers.

Every organization’s approach  will vary based on how they run their business, says Amit Patel, a senior vice president at Consulting Solutions. “Businesses need to consider their budget, risk tolerance, the strengths and weaknesses of their creative staff, and the roadblocks in their current creative processes,” he says.

Collaboration with business, technology, and creative managers is essential for making decisions such as whether to invest in building a custom model to yield more precise results and for identifying the use cases that fit their budget, risk level, corporate culture, and creative processes. Some companies also have an AI council with stakeholders from areas such as legal and compliance, which performs a value, risk, and ethical analysis of the tools.

Getting started

AI is quickly being embedded in businesses’ creative processes and will soon be a function of every digital creative tool. The most effective managers will move beyond skepticism or fear to an awareness of how AI can help to produce higher quality creative work faster and more cost effectively.

“We won an award for the best use of emerging technology when we created a song [using AI], crafting our own lyrics and letting the AI handle the music,” says Omnisend’s Meyer. “For product releases, we’ve turned to a text-to-voice tool, which allows our product managers to turn their written scripts into clear, engaging voiceovers, which helps us communicate our product updates with ease.”

“We are now entering the next phase of integrating AI into marketing and enterprise-wide processes,” says Everest Group’s Verma. “Given the rapid development of new AI options, this is not a one-off but a continuous learning process. Organizations and managers must encourage creative professionals to learn about AI, to experiment with it, to become familiar with it.”

Read more