How AI Could Save $4 Billion (and likely more): Intro to AI Powered Content Marketing
Discover how AI is transforming content marketing, offering a potential $4 billion savings. This article delves into AI's role in creating and optimizing content, from generating text and visuals to analyzing vast data. Learn how AI complements human creativity in marketing, with insights on bias, accuracy, and the balance between technology and human touch.
Artificial Intelligence (AI) is the next frontier in marketing. It will revolutionize the field by 2030. AI is ready now to: help author short and long-form content; generate images and video to support the narrative; and create music to supplement the storytelling. AI can help marketers brainstorm ideas as well as leverage surveillance through rapidly analyzing social media and other types of data to inform content strategies. AI can optimize the distribution of messages, personalized ads, and content across various channels. Marketers need to understand how advancing technology will alter and automate some of the common aspects of marketing and advertising that we take for granted today.
Modern advertising has only been around for about 100 years; its overarching goal is to: deliver relevant messages to consumers in an efficient and timely manner to drive action. That broad mission will not change, but how we do that will. Content marketers who embrace AI now will be ahead of the competition and reap the most rewards in the future. To be clear, AI's role is to help support the narrative we are creating. It is not to solely produce it. I would never advocate or advance the idea that a marketer should simply enter a request of an AI, hit the return key, and take its output as a piece of text to be copied and pasted directly into a campaign. In this article, we'll explore how AI can enhance your content marketing efforts and help you reach your target audience more effectively.
There are many kinds of AI, but for our purposes, we will focus primarily on Generative AI, which generates new content based on specific inputs using various methods, such as machine learning algorithms or neural networks. Generative AI produces a variety of outputs spanning images, videos, audio, or text. While software developers may interact with Generative AI capabilities by APIs or coding, non-technical (or technical) professionals can use some of these applications by inputting simple textual prompts. For example, a user who wants to create an image could type in "a dog riding a skateboard" and receive an output of such an image. Or a user who wants to create a story of a dog riding a skateboard could use Generative AI to write unique content. Developing and refining these textual inputs is called prompt engineering. Generative AI increases the efficiency and productivity of marketers by putting the power of AI at their fingertips. However, while Generative AI can help to streamline content creation, it may not always produce high-quality content that effectively engages and resonates with audiences. Marketers should be mindful of this and consider using a combination of both human and AI-generated content in their marketing strategies. Therefore, Generative AI is just another tool in a marketer's toolbox.
Tools alone do not make masterpieces, nor does a novice with fancy tools. Having the right tools is only one part of developing an effective marketing campaign. As with any tool, it takes skill, knowledge, and experience to use AI effectively. AI can be helpful, but it’s essential to ensure that the content produced is accurate and aligns with the overall goals and values of the brand. Therefore, marketers should not use AI just to say they used AI. Like any other tool in the marketer’s toolbox, it must align with the mission they are seeking to accomplish.
Heinz expertly demonstrated how to appropriately leverage AI in a recent campaign. Heinz introduced their tomato ketchup in 1876. Through more than a hundred years of advertising, Heinz holds 60% of the market in the US and 80% in Europe.[1] Heinz capitalizes on its market penetration in its advertising. In a 2021 commercial, Heinz asked people and kids across the globe to “draw ketchup.”[2] Bathed in decades of advertising, it is not surprising that the iconic design of the Heinz ketchup bottle is what came to mind. One person said, “It’s the only ketchup I know.” So, part of the Heinz brand story is that they are synonymous with ketchup. In 2022, they used an AI model that generates images (Dalle-2) to take this brand story to the next level: draw ketchup. Like the people in their prior campaign, the AI consistently drew images that looked like a Heinz ketchup bottle – from a Renaissance-inspired version to stained glass; all images reflected the iconic brand’s image, allowing the commercial to conclude: "Even AI knows that Heinz is ketchup."[3] Heinz consistently told their brand story; AI enabled a clever spin on their overall message.
This example also highlights the implicit bias within AI: it only knows what it has been trained on. The Kraft-Heinz corporation spends over a billion dollars annually on marketing its products.[4] Therefore, it isn't surprising that AI capabilities that have largely been trained on content available on the internet would consume an over-representation of Heinz's distinctive tomato ketchup. There are, in fact, many brands of ketchup and many types of ketchup, such as ones based on mushrooms, fish, or fruit. Therefore, while AI seems dazzling in its generation, we must remember it is only as good as the data it has been trained on.
Consequently, AI may overrepresent the opinions and images of majority groups who have had access to publicizing their beliefs in media and underrepresent those of minorities. While this naturally brings up thoughts of how men, Americans, and Europeans have historically been archived more than women and racial/ethnic minorities, the bias goes further than that, it goes beyond bias to a lack of creativity – AI does not think, nor is it creative: it regurgitates. For example, I have run hundreds of scenarios OpenAI’s Dalle-2[5] where I have asked AI to draw a unicorn. In nearly all scenarios, the AI makes it white regardless of the description or context I give in my instructions. If I ask for a unicorn in a different color, it will comply, but without that request, I have learned all unicorns are white. But how could this be true of a mythical creature? Because the bias in representation goes far beyond skin tones and gender. Historical bias influenced the construction of many elements we take for granted today – such as, all unicorns are white. Historically, the concept of a unicorn existed across cultures and societies and unicorns came in many hues. But now they are white. Additionally, I have also run hundreds of scenarios asking the AI to create an umbrella using Midjourney.[6] In all scenarios, the umbrella is open – I have never received a closed umbrella (or one in any other state). How is this possible? People do not keep their umbrellas open; umbrellas spend most of their existence closed, but in art and in online advertising they are more frequently depicted open. When we think of bias or lack of diversity, we often think in racial or ethnic terms. Bias and a lack of diversity can be present in many subjects, like umbrellas. So, keep this in mind when your fingers are typing into an AI to create incredible images, text, etc., that the product only will produce whatever preprogrammed bias exists. If you are not conscious of this, you may perpetuate and amplify representation voids.
Now that we are grounded enough in AI basics let's discuss its implications for content marketing. From social media posts, emails, videos, podcasts, webinars, and infographics, the number of words and images compiled yearly by companies to advertise their products is staggering. Stock photography alone accounts for over $4 billion[7]. Add in the rates for copywriters, graphic designers, musicians, agencies, etc., and it is no wonder that globally over a trillion dollars is spent on marketing.[8]
In the US, over $400 billion is spent annually on content marketing[9]– if the promise of AI brought only a 1% decrease in the overall content marketing spend in the US, that would be a $4 billion savings. The low cost of AI will undoubtedly tantalize chief financial officers; at the same time, it may be driving fear into the hearts of chief marketing officers because of thoughts of job losses or threats to creativity.
However, AI is not intended or able to replace humans. Instead, it is an automated tool to help marketing teams be more thoughtful and efficient. AI can also automate mundane, process-oriented tasks, allowing humans to focus on more creative tasks. AI can also assist marketers, allowing them to spend more time curating content and less time on repetitive tasks. AI can enhance the content marketing process but cannot replace the human experience and creativity that is the foundation of all good advertising and marketing.
In the creative process, the first step AI can help marketers with is the analysis of massive amounts of data. More pictures are posted on Instagram daily than were taken in the entire nineteenth century. Every minute there are nearly 1,000,000 posts/comments/replies made on Facebook.[10] A human would fail at attempting to detect trends or patterns in this dizzying amount of data. Researchers have used Twitter data for several years to track and predict the spread of infectious diseases (Ebola, influenza, COVID-19) and noncommunicable diseases (heart disease, depression).[11] While the golden age of Twitter research has likely come to an end, AI can also mine Reddit, Quora, online news, etc. Furthermore, by using these real-time sources and historical data, marketers can then use this data to anticipate customer desires and craft campaigns that are more likely to be successful.
After analyzing data to understand your target audience and identifying trends, the next step in the content marketing process is often brainstorming and prototyping ideas. The conceptualization process can be time-consuming and resource-intensive, especially if you are working with a large team or trying to target multiple segments of your audience. However, AI can help streamline this process by quickly outlining campaign pitches or creating mockup storyboards. For example, an AI tool could generate various ideas and concepts based on the company's target audience and desired outcomes, allowing the marketing team to review and evaluate potential options quickly.
Once the campaign brief is finalized, it's time to start creating the text, images, and sound that will bring it to life. Depending on the audience, the type of content needed may be short form, medium form, or in-depth. AI's role in producing this type of content will vary. Let's start with a content matrix framework to help determine how best to use AI for textual content production. In my experimentation, AI is best for writing concise statements. When I have it elaborate on a topic, it reiterates the same point through multiple sentences – something a good writer would never do. Therefore, AI is most useful for creating short content in highly engaging mediums such as social media or digital advertising.
As the marketer draws their consumer in, they may need to engage that consumer on a web page or blog. This content will need to be longer and may not be as engaging or as pithy as social media posts. Campaigns sometimes need longer form content to educate or inspire its target customer. Here I recommend leveraging AI to create an outline or initial framing of the article. Long-form content takes time and a skilled author to ensure it is engaging and adequately educating the customer.
Lastly, for in-depth content that could power a webinar or be a research paper, AI's potential help to the author is limited. It's not that AI can't write thousands of words that interweave into a compelling article; it is that AI sometimes lies. I have found that in my many queries with AI, it always answers my questions or produces the content I've asked for; but it needs to be checked for veracity. Ask AI for an essay that includes scientific citations and URLs, and it will produce it. Although they will appear very compelling, the citations are likely fake, and the URLs are made-up. For example, using the paid version of GPT-3 and using the Davinci model, which is their best and most expensive, I asked AI to: “write an article about why people should drink water and include citations with URLs from scientific journals.” Here are the references:
· Centers for Disease Control and Prevention. “How Much Water Should You Drink?” https://www.cdc.gov/nutrition/data-statistics/plain-water-the-healthier-choice.html
· Toshinai, K., Nakata, H., and Tsutsumi, K. “Water-Induced Thermogenesis”. https://www.mdpi.com/2072-6643/3/8/1254
· Anurag Ramkrishnan, Ashok Erramilli, and Hemant Kasliwal. “The Role of Water in Health Maintenance and Disease Prevention”. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4799531
The first article is real. The second returns a 404 file not found error. The third looks so compelling, but it does not exist. The article the link actually is to: “Systematic engineering of the central metabolism in Escherichia coli for effective production of n-butanol” by Mukesh Saini, Si-Yu Li, Ze Win Wang, Chung-Jen Chiang, and Yun-Peng.[12] The word “water does not even appear in this article. Having spent so many hours in dusty, poorly lit libraries pouring over microfiche, my heart skips a beat whenever I ask AI to provide scientific citations and I watch them magically appear on the screen. But magic isn’t real. The ability to accurately cite its sources is not only a limitation for scientific information, it also does not allow those whose seminal ideas are influencing its output. AI can also write code for coders and here too there is a lack of attribution. While there are many open-source code available, some of it requests attribution. Perhaps these are just a limitation of the AI capabilities available today, but for now, when writing in-depth content, AI collaboration with the copywriter will be minimal.
Now that we have completed our brainstorming and content writing, it is time to create an image or video that will support our storytelling. When using AI for image generation, a novice artist might find it hard to translate the idea in their mind onto the screen because the novice needs to understand the underlying structure and form of artwork. Art combines many factors, including composition, line, color, expression, light, and texture. A novice does not possess the vocabulary to translate the vision in their mind to text. It is foolish to think that AI will be the demise of commercial artists. A seasoned artist will gain superpowers when using AI. They can generate mock-ups or a base drawing for them to build upon. The graphic artist will still need a graphics program and the skills to use it to refine the image. Like the copywriter, the graphic designer will collaborate with the AI capability. Through the intellectual property the artist is imbuing the base AI image with; it truly transforms the image from a simple AI-generated image.
The legal landscape on AI copyright protection undoubtedly will take years to truly be settled. Historically, the US Copywrite Office has denied solely AI generated art a copyright because they protect: “the fruits of intellectual labor [that] are founded in the creative powers of the [human] mind.”[13] In September 2022, an AI artist announced they had received (perhaps the first) copyright for their art.[14] But then several weeks later, the US Copywrite Office reversed their decision.[15] I expect this to be an evolving topic over the next few years.
A marketing campaign may also require music, and AI can help with this too. First, we need to decide whether we want to create our own new music or use AI to generate something in the style of someone from the past. In 1827, Ludwig van Beethoven died. Many remember Beethoven for his powerful 9th Symphony, also known as the "Ode to Joy." But he was actually commissioned to write two symphonies: the ninth and the tenth. While he did start the tenth, he died before its completion. In 2021, an AI capability, coupled with an expert musicologist, used the written notes and musical sketches left by Beethoven to complete this work[16]. It debuted 251 years after his birth and can be streamed online[17]. If you are looking for something more modern, The Lost Tapes of the 27 Club brings awareness to mental illness by using AI to create new music by artists such as Jim Morrison and Amy Winehouse, who tragically all died at the age of 27.[18] For something a little less morbid, Sony has also used AI to create a new Beatles song.[19]
Musicians can also use AI to collaborate with it to create their music. If you think copyright might be a concern, AI has already developed every possible pop melody that could be composed – all 68 billion of them.[20] Pop melodies are made up of a finite range of musical notes: “eight notes up, major and minor, and twelve notes across.”[21] They are also the only type of music that gets litigated because they can make money. In 1976, George Harrison (formerly of the Beatles), was found guilty of "subconsciously" plagiarizing the 1962 hit “He's So Fine" for his 1970 song "My Sweet Lord." The judge stated that Harrison subconscious mind: “knew this combination of sounds would work because it already has worked in a song his conscious mind did not remember.”[22] This bunch of psycho mumbo jumbo cost Harrison over $500k and the issue continued to be litigated for the next two decades. To eradicate lawsuits like this, Damien Riehl and Noah Rubi used AI to create all the possible pop melody combinations. Additionally, Peter Burk generated all chord projections that can occur over the seven musical octaves.[23] Both of these datasets, the melodies and the chords, are publicly available for others to continue to expand upon. Therefore, modern composers may use AI with limited fear of lawsuits.
As musical artists put their fingers on a computer keyboard, rather than a piano keyboard or guitar bridge, they can type in a few notes or chords and work with AI to develop their orchestration further. Musicians may also use AI capabilities to add tracks of various musical instruments to their scores. So, for a marketing campaign, AI could help create a catchy jingle, the background music to a commercial, or a pleasant score in a webinar.
Now that we have completed the marketing campaign elements, it's time to distribute it through email, search engines, and social media networks. AI algorithms can analyze data from search engines and social media platforms to identify patterns and trends that can be used to optimize campaigns and target specific audiences. Furthermore, with real-time insights into campaign performance, marketers can quickly make changes and tweak campaigns, providing a more dynamic, responsive marketing approach. In addition, through automation, AI can constantly optimize campaigns and personalize the ads that are shown to consumers who are more likely to engage and convert users. Finally, by leveraging AI technologies and data analytics, marketers can maximize their ROI from marketing campaigns, ensuring their limited resources don't go to waste.
I once asked AI how it thought marketers should collaborate with it and what role, as humans, they would play in this brave new world of AI-powered content marketing. It said, "The content marketer's job will be to be a human interface between the company and its AI content creators."[24] I agree with this assessment. We interact and interface with technology every day. I do not fear AI or think it will be the death knell for content creators. Job responsibilities change and adapt to new technology. A few decades ago, marketing teams had people who manually mimeographed copies of materials for mass distribution. Then they got photocopiers. Change happens - how we adapt to it is of the most importance. Adapting to change involves thinking quickly, making educated decisions, and being open to new experiences. It is important to remember that while change can be difficult, it can bring new opportunities. I have seen a lot of changes in my life, I have found it is beneficial to remain optimistic, focus on the potential benefits of the change, and see how I can be part of it. Therefore, astute marketers or students looking to enter marketing must embrace the opportunities AI offers for content marketing. Through our active collaboration with it, we can rapidly analyze data; brainstorm ideas; create content spanning text, images, video, and music; and optimize its distribution. The marketing leaders of tomorrow will be forged during this AI marketing evolution. I hope you are one of them.
References
1. QZ Ketchup. Quartz, 2017. Available from: https://qz.com/emails/quartz-obsession/1084956/ketchup
2. Heinz Ketchup Canada, Heinz Draw Ketchup. 2021: YouTube Available from: https://www.youtube.com/watch?v=APoGHH1Ns2M.
3. Heinz, Heinz A.I. Ketchup. 2022: YouTube Available from: https://www.youtube.com/watch?v=LFmpVy6eGXs.
4. Naidu, R. Kraft Heinz to Increase Marketing, Sees Supply Chain Savings of $2 Bln by 2024. Reuters, 2020. Available from: https://www.reuters.com/article/kraft-heinz-strategy/kraft-heinz-to-increase-marketing-sees-supply-chain-savings-of-2-bln-by-2024-idUSL4N2GB414
5. OpenAI. Dall-E 2. Available from: https://openai.com/dall-e-2/
6. Midjourney. Midjourney; Available from: https://www.midjourney.com.
7. Stock Images and Videos Market - Global Outlook & Forecast 2022-2027. 2022, Arizton Advisory and Intelligence Available from: https://www.arizton.com/market-reports/stock-images-and-stock-videos-market.
8. Bauer, T., J. Gordon, and D. Spillecke The Dawn of Marketing’s Next Golden Age: $200 Billion and Counting. 2023. Available from: https://www.mckinsey.com/~/media/mckinsey/business%20functions/marketing%20and%20sales/our%20insights/is%20sports%20sponsorship%20worth%20it/marketing%20golden%20age.pdf
9. Absolute Reports Pvt Ltd Content Marketing Market Growth 2022-2028 | Growth Estimation, Major Manufacturers, Industry Share, Regional Analysis, Types, Applications, Opportunities, Challenges, Drivers, Trends. GlobeNewswire, 2022. Available from: https://www.globenewswire.com/news-release/2022/04/18/2423491/0/en/Content-Marketing-Market-Growth-2022-2028-Growth-Estimation-Major-Manufacturers-Industry-Share-Regional-Analysis-Types-Applications-Opportunities-Challenges-Drivers-Trends.html
10. Osman, M. Wild and Interesting Facebook Statistics and Facts. Kinsta, 2023. Available from: https://kinsta.com/blog/facebook-statistics/#:~:text=Of%20course%2C%20in%20that%20one,and%20136%2C000%20photos%20are%20uploaded
11. Debogovich, G. Tweets, Posts, and Searches: Digital Surveillance Is the Next Horizon in Healthcare Analytics. Medium, 2023. Available from: https://medium.com/@ginadebogovich/tweets-posts-and-searches-digital-surveillance-is-the-next-horizon-in-healthcare-analytics-ce8386799c83
12. Saini, M., et al., Systematic Engineering of the Central Metabolism in Escherichia Coli for Effective Production of N-Butanol. Biotechnol Biofuels, 2016. 9: p. 69. Available from: https://www.ncbi.nlm.nih.gov/pubmed/26997975
13. U.S. Copyright Office Review Board, Second Request for Reconsideration for Refusal to Register a Recent Entrance to Paradise (Correspondence Id 1-3zpc6c3; Sr # 1-7100387071) U.S. Copyright Office Review Board, Editor. 2022: Washington, DC Available from: https://www.copyright.gov/rulings-filings/review-board/docs/a-recent-entrance-to-paradise.pdf.
14. Edwards, B. Artist Receives First Known Us Copyright Registration for Latent Diffusion Ai Art. Ars Technica, 2022. Available from: https://arstechnica.com/information-technology/2022/09/artist-receives-first-known-us-copyright-registration-for-generative-ai-art/#:~:text=Despite%20popular%20misconception%20(explained%20in,author%20instead%20of%20a%20human
15. Artist Fights for Copyright for Ai-Assisted Graphic Novel (Correct). Bloomberg Law, 2022. Available from: https://news.bloomberglaw.com/ip-law/artist-contests-copyright-denial-for-ai-assisted-graphic-novel
16. Elgammal, A., How Artificial Intelligence Completed Beethoven’s Unfinished Tenth Symphony, in Smithsonian Magazine. 2021 Available from: https://www.smithsonianmag.com/innovation/how-artificial-intelligence-completed-beethovens-unfinished-10th-symphony-180978753/.
17. Sam, Beethoven X: The Ai Project: Complete (Bonn Orchestra) 2021 Available from: https://www.youtube.com/watch?v=Rvj3Oblscqw&t=4s.
18. Lost Tapes of the 27 Club: Using Ai to Create the Album Lost to Music's Mental Health Crisis. 2021 [cited 2023 Jan 8]; Available from: https://losttapesofthe27club.com/.
19. Sony CSL, Daddy's Car: A Song Composed with Artificial Intelligence - in the Style of the Beatles. 2017 Available from: https://www.youtube.com/watch?v=LSHZ_b05W7o.
20. Madrigal, A.C. The Hard Drive with 68 Billion Melodies. The Atlantic, 2020. Available from: https://www.theatlantic.com/technology/archive/2020/02/whats-the-point-of-writing-every-possible-melody/607120/
21. Riehl, D., Copyrighting All the Melodies to Avoid Accidental Infringement, in TEDx Talks. 2020 Available from: https://www.youtube.com/watch?v=sJtm0MoOgiU&t=902s.
22. George Harrison Guilty of Plagiarizing Subconsciously, a '62 Tune for a '70 Hit, in The New York Times. 1976. p. 42 Available from: https://timesmachine.nytimes.com/timesmachine/1976/09/08/79684610.html?pageNumber=42.
23. Burkimsher, P., Chord Progressions, in GitHub. 2017 Available from: https://peterburk.github.io/chordProgressions/index.html.
24. Open AI. Openai Api. Available from: https://openai.com/api/.
