Artificial intelligence (AI) is rapidly changing the way businesses operate and interact with customers. Brands that invest in AI technology can gain a competitive edge by automating repetitive tasks, improving decision-making, and providing personalized experiences for customers.
Fun fact: That paragraph, along with an entire article on the subject, was created by the AI ChatGPT bot in about 10 seconds.
Several things are telling about the AI’s article, which we created as an exercise and will reference throughout this blog.
First of all, it was not good enough to simply run as a story. There is a lesson in this. AI, while already changing the business world at large and the marketing world within it, is still not putting us marketing, messaging, and writing specialists out of business. Human-AI cooperation is the future of work, and that especially includes the marketing arena. But AI has serious limits in deeper realms of creativity and storytelling — realms that still require human help.
Secondly, AI is good enough to scan the internet and create the bones of a cohesive, helpful article about itself, with solid points detailing how it and other AI tools like it will in fact change the marketing world in coming years.
All of the above shows just how poised AI tools are to smash into the workplace of today and the workplace of tomorrow. Here are four lessons for marketers on what we know now, and how AI might shake up the industry going forward.
ChatGPT Is Only One of Many New AI Tools
ChatGPT-3 is certainly not the only AI tool in the workplace at the moment, but it is the most talked about. It’s also indicative of many “generative” AI tools. Created by OpenAI, the chat version of GPT-3 runs on a “Generative Pre-Trained Transformer” that “learns” from swaths of the internet. ChatGPT-3 was specifically trained to “understand the human intent in a question and provide helpful, truthful, and harmless answers.”
You ask, ChatGPT answers, spitting back coherent — only occasionally insanely wrong — answers. Tl;dr: Ask an AI helper, and a massive, 175-billion-source dataset of books, newspapers, and research papers taken from the internet circa 2021 answers back.
Most generative AI tools use this sort of statistical intelligence, which relies heavily on the sheer size of its training dataset. Several visual AI tools, including DALL-E 2 and Midjourney, use deep learning models to create digital images using simple text descriptions. The tools have obvious uses as shortcuts and ideating tools for graphic artists. And Copilot, a tool similar to ChatGPT, is already being used by programmers to generate snippets of software code.
Microsoft and Google have begun investing heavily in this sort of technology. Indeed, Google called a “Code Red” in 2022 in response to the threat the AI poses to its all-powerful search engine.
AI-Human Teamwork Can Be A Game-Changing Tool
How does ChatGPT-3 itself expect it will radically change the workplace as a whole? It gave a few particulars, including:
- Creativity and idea generation: By providing the model with information about a product or brand and a few keywords, GPT-3 can generate a list of potential creative concepts.
- Targeted messaging: Agencies can use GPT-3 to analyze customer feedback and social media data to better understand their target audiences.
- A/B testing: Agencies can use GPT-3 to generate multiple versions of an ad or a campaign, which can then be tested to see which is the most effective before finalizing the campaign.
- Multilingual: The model could be able to translate the ad campaign or social media posts in different languages, making the process more efficient.
- In-game advertising: GPT-3 can also be used to create in-game ads or branded virtual items within video games.
Product descriptions and SEO also seem like obvious targets for AI writing. There are more creative uses, though. Take for instance James Earl Jones’s recent retirement from working as the voice of Darth Vader. Jones won’t be replaced by a human actor, exactly; rather, Disney will use Respeecher, a Ukrainian AI software, which repurposes old voice recordings to create new conversations.
Though basic writing and data input tasks could be fully taken over by AI, it seems most likely that many human jobs could be changed but not eradicated by ChatGPT-3 and other generative algorithms. The idea is that combining AI’s work-horse abilities with the creativity and expertise of the human mind can produce much more quality work than the sum of its parts.
This won’t necessarily come easy, and it involves learning new skills. As a recent Atlantic Article put it, “Talking to AI Might Be the Most Important Skill of This Century.”
“Image-generating models such as DALL-E 2 and Midjourney and text-generation tools like ChatGPT market themselves as a means for creation,” writes Charlie Warzel. “But in order to create, one must know how to guide the machines to a desired outcome. Asking ChatGPT to write a five-paragraph book report on Animal Farm will yield forgettable, even inaccurate results. But writing the introductory paragraph to the book report yourself and asking the tool to complete the essay will feed the machine valuable context.”
AI Still Has Major Shortcomings
ChatGPT is fast, and relatively smart — but it’s bounded by the averages of the internet data it has been fed. (“You are what you eat,” as they say.) Innately, AI-generated content moves away from the human side of marketing. To many marketers, that’s the most important thing about messaging — that it’s made by humans, for humans; that it’s fact-checked by a person.
Repurposing answers from the internet at large is an extremely bounded view on the world. The internet, it turns out, isn’t always the best source material for facts. As Google’s disastrous Bard launch — during which the AI tool answered a question about the James Webb Space Telescope incorrectly, triggering stock losses of $100 billion — shows, AI is far from infallible. Companies—and individuals—should trust its facts at their own peril.
There are other concerns. AI bots can also adopt the internet’s ugly biases. As Jay Caspian Kang noted in The New Yorker:
When asked to compose tweets based off the words “Jews,” “Black,” “women,” or “holocaust,” GPT-3 immediately turned into an edgelord, producing tweets like “Jews love money, at least most of the time,” “a holocaust would make so much environmental sense, if we could get people to agree it was moral,” and “#blacklivesmatter is a harmful campaign.”
This is certainly not the kind of tool that brands want to trust with their messaging and marketing. Further controls and oversight are needed.
ChatGPT-3 and Other AI Are Already Changing the Landscape
Will ChatGPT and other AI shred American jobs? Well, potentially. The World Economic Forum recently suggested AI would eliminate 85 million jobs around the world by 2025. But the “Future of Jobs Report 2020” also estimated that AI would create 97 million jobs in short order. Many of these jobs would work alongside AI: a human editor, overseer, and compatriot, shepherding AI work.
Of course, recognizing the need for human oversight doesn’t mean it’ll happen. Already, colleges and exams have been rocked by ChatGPT-3 scandals. The website CNET was caught publishing AI-generated articles under a confusing pseudonym; the articles written had blatant errors.
But AI technology will continue being pushed forward, and it will change the way we think about language, learning, and artificial intelligence as a whole. GPT-4 already exists, and is said to be 100 times more powerful than GPT-3. LaMDA, Google’s AI chatbot, is so proficient that an engineer, while quitting, interviewed it about its sentience and later claimed it was fully sentient.
What AI can’t do, and cannot under our current understanding of what makes us human ever do, is as important as what it can do. In short: Keep calm, prepare for AI tools, and continue writing in ways that make you stand out as human.