Over the past several months, your LinkedIn feed has probably been flooded with posts about ChatGPT and generative AI—and considering that the topic has been addressed in popular media from local news networks to South Park, it’s safe to say this trend has gone mainstream.
Generative artificial intelligence creates nearly instant content: it receives input from the user, references an immense database, and in a matter of seconds or minutes (depending on the complexity of the prompt) returns text or images based on the relevant information it referenced. Marketers can direct these tools to create content in several formats, and you can even prescribe word counts, communication objective and tone. The result is a human-like (albeit generic) rehash of aggregate available content.
The prospect of creating content en masse for (basically) free is an exciting one—and a 2022 Oracle study on marketing trends indicates that marketing leaders agree:
- 42% of marketers already trust AI to personalize content and offers in real time.
- 30% of marketers fear they are missing out on AI’s capability to predict behavior and adjust campaigns.
- 21% of marketers trust AI to write copy.
These numbers are likely to trend upward for a while. Research around generative AI chatbots (like ChatGPT) increased 13% from Q4 2022 to Q1 2023, and 20% of marketing and advertising executives believe AI will be “critical” to their job functions by 2025. Generative AI is part of the marketing tool stack now, and it’s important to understand its potential uses and limitations.
How marketers are using AI in ecommerce
For the most part, marketers are putting AI to work on simple, repetitive tasks that would otherwise involve hours (or even weeks) of human labor. Per both the Oracle study and a 2022 Ascend2 report, most of the areas in which marketers trust AI to do the work for them concern targeting and optimization, with a few task types that involve creating content that the customer engages with.
Example treatments include optimizing email send times, ad placement, segmentation and product recommendations. AI is also being used to calculate conversion probability—which can be helpful when crafting promotional email campaigns, social media ads and product detail copy.
Additionally, marketers are putting generative AI to work creating content at scale. For example, in Q1 of 2022, 32% of marketers were either using or testing AI to personalize email messages and offers, and 22% were using it to personalize email subject lines. ChatGPT has been put to work creating product descriptions as well as copy and images for paid ads—which can be especially useful for multivariate testing across large audiences. And of course, marketers have found chatbots specifically useful for handling live chats with website visitors.
Furthermore, generative AI is a remarkable tool for ideation and brainstorming. Using ChatGPT to produce starter product copy and headline suggestions can be a huge timesaver for marketers today.
Current limitations of generative AI for marketers
Artificial intelligence tools are naturally limited by the data on which they are trained. ChatGPT can’t always discern good source material from suspect source material, and it’s even known for “hallucinating” completely made-up ideas. This means any content generated by AI must be vetted by humans for common-sense editing and fact checking.
Marketers using generative AI also have to deal with the “black box” problem. Depending on which generative AI you’re using, you may be unable to tell specifically where it sources its information, how it evaluates that information or why it makes the decisions it makes. This can make the editing and review processes difficult for humans, because while you can ask a fellow human for their rationale, having this conversation with a chatbot can at times be maddeningly tedious.
Perhaps the biggest limitation of generative AI is the generic nature of its results. While ChatGPT can accept prompts for tone and voice, the outputs are ultimately dependent on how ChatGPT interprets those prompts.
They tend to read like the sorts of things anyone could have written, so you won’t find much in the way of novelty or creative insight here. ChatGPT can create content that reads like a human made it, but that isn’t the same as content that reads like a human expert made it. For some marketing assets, this is fine—but you probably don’t want ChatGPT to write high-profile assets just yet.
One final note on limitations: AI is a tool, and like any tool, its effectiveness comes down to the people wielding it. The advent of generative AI has led to a host of “overnight experts,” but marketers are still figuring out how to best use these tools.
If used judiciously, AI has the potential to revolutionize brand campaign planning with powerful tools for data analysis, personalization, automation and customer engagement. And like most trending technologies, marketers who harness the capabilities of AI effectively have the opportunity to gain advantages in efficiency and campaign efficacy over their competitors.
To learn more about how PriceSpider is helping brands grow with cutting-edge technology, see how we’re giving control of the buyer journey back to our customers and making every moment shoppable.