Chatbots and image generators are the latest technology piquing marketers’ interest.
It was only a matter of time before advertisers got their hands on AI. To be fair, the technology has been used in ad tech for years, but generative AI, a subtype of artificial intelligence that has obvious—and fun—applications, has instantly changed how marketers are preparing for adoption.
Bots like ChatGPT and DALL-E 2 have exploded into the mainstream, and industry watchers are predicting the tech will soon disrupt nearly every stage of advertising, from brainstorming to copywriting to targeting ads. But there’s a lot more to be said for the tech, including its current use cases, its shortcomings and its downright dangers.
To catch you up to speed, here’s everything you need to know about generative AI.
What Is Generative AI?
Generative artificial intelligence is a form of AI that, as its name suggests, generates content. Depending on the specific platform’s capabilities, this content can be images, videos, text, audio as well as other mediums. The platforms that have recently become popular generate their content from a user’s input, such as “Produce an image of X” or “What is the answer to Y?”
How Does Generative AI Work?
The simplified answer is that most generative AI models work by pre-training their neural networks—essentially, their brains—on large datasets, culled from books, websites, articles and other sources. The networks are able to learn patterns and features of the data, and from that knowledge create new data.
As for a model that produces content from a user’s query, that model also undergoes a process called fine-tuning, in which it runs the data of the query through its neural network and adapts its pre-trained knowledge to generate a response specific to the query.
What Are the Top Generative AI Platforms?
Generative AI platforms first pierced mainstream circles with the limited release of DALL-E 2 in April of last year (“DALL-E” is a portmanteau of painter Salvador Dalí’s name and that of the animated Pixar robot, WALL-E).
Its popularity exploded in the fall when the platform became fully open to the public. DALL-E 2, the more advanced sequel to DALL-E (released in January 2021), produces images based on an input of text. The bot is capable of producing content in a range of styles, and based on virtually any query. For example, from the query “A raccoon playing tennis at Wimbledon in the 1990s,” DALL-E created the image below:
A raccoon playing tennis at Wimbledon in the 1990s #dalle2 #dalle pic.twitter.com/45tBFKJ9ZJ
— Best Dalle2 Pics (@Dalle2Pics) May 14, 2022
At the moment, the most popular platform is ChatGPT, which rolled out to the public at the end of last year. ChatGPT, like DALL-E, takes text-based inputs, but differs in that it generates text-based outputs, as opposed to images. Its content can be anything from answers to rudimentary questions (e.g., “What is the tallest building in the world?”) to explanations of more complicated queries (e.g., “Explain how a developer could create his own crypto exchange from scratch.”). It can even produce novel material based on highly specific entries (e.g., “Write a song about the fall of Rome, in the style of Taylor Swift.”).
In essence, the bot has been described as a super-capable search engine that can provide clear, instant and humanlike responses for a wide range of queries.
Other popular generative AI platforms include Stable Diffusion and Midjourney, which are both image-to-text generators like DALL-E. And maybe you’ve seen the bots that produce human faces that don’t actually exist? Those are generative AI, as well.
🦾 Related Reading: Embracing AI — 5 Reasons Marketers Shouldn’t Fear Artificial Intelligence
How Can Marketers Use Generative AI Like ChatGPT for Marketing?
Marketers are already experimenting with generative AI platforms to see how they could benefit their business. Perhaps the most common use case thus far has been idea generation, which, while abstract, can still inform a brand’s marketing materials, and at a much faster clip than it would normally take. For example, if an agency is in the brainstorming stage for a new campaign, it can plug relevant queries into DALL-E 2 and see what the platform spits out. The same goes for ChatGPT, which can outline ad copy for just about any concept.
Some marketers are already using these models to create ready-to-go advertisements. Canadian agency Rethink ran a campaign last year highlighting images of Ketchup generated by DALL-E 2. Earlier this month, Ryan Reynolds debuted an ad for his wireless brand Mint Mobile that was partially penned by ChatGPT. Going forward, marketers expect new and increasingly concrete applications to become available as generative AI develops.
You knew it was just a matter of time until we did this (extend the @MintMobile savings with @OpenAI, that is). pic.twitter.com/uf2jblpG2j
— Ryan Reynolds (@VancityReynolds) January 10, 2023
Are There Any Negative Side-Effects of Using Generative AI?
The most pressing concern of generative AI is with misinformation. Since these models are only as good as the data they are trained on, if that data is false or biased or somehow corrupted, then their generated content will be so as well.
In fact, OpenAI, the startup behind ChatGPT and DALL-E, admits that while ChatGPT’s answers may sound plausible, some may in fact be completely wrong. Stack Overflow, a forum for programming-related questions, has already banned users from posting answers created by ChatGPT because of the model’s low rate of accuracy. That ChatGPT’s answers look and sound correct makes them that much harder to falsify.
AI platforms are also not necessarily up to date on the facts. ChatGPT, for example, is limited to knowledge of 2021 data. When queried about crypto firm FTX—which collapsed last fall—the model still describes it as one of the most popular exchanges, as well as having high liquidity.
Hardees recently ran a campaign, from Dubai agency And Us, that showed off some of the limitations of using AI for creative:
Issues of plagiarism are another concern, especially with regard to image generators. All of the data the models have been trained on comes from somewhere and someone, and without knowing it, an agency could create images that directly crib the style of an artist. This is why copyright will likely play a sizable role in the future of AI technologies.
Finally, and with special significance to marketers, generative AI could open new questions of data privacy. Technologists are already proclaiming how AI will disrupt targeted advertising once companies can upload their data to a model’s neural network. But how will consumers feel that a highly intelligent computer knows all kinds of information about them and can create an unlimited amount of novel content from that information, some of which may very likely be manipulative? These questions and more will be explored as AI develops.
Will Generative AI Bots Replace Creatives?
Fear not, copywriters, because ad experts say that generative AI bots will not replace you. Issues of taste, finesse and all other refinements that make copy great are still very much lacking in generative AI, which is why they are for now being used mainly as a springboard for inspiration.
That said, copywriters who know how to use generative AI bots could replace you. This is why ad execs are calling on the industry to ramp up engagement with these tools, so that the combined power of humans and robots can provide the best possible outcome.
But who knows where AI will be in five, 10, 15 years down the road? Is it so flabbergasting to imagine an AI bot that can learn taste as it can the names of the planets? I guess we’ll find out.
This article was written by Asa Hiken from Ad Age and was legally licensed through the Industry Dive Content Marketplace. Please direct all licensing questions to [email protected].
