8 analyst predictions on the future of ChatGPT and generative AI in marketing and retail

“[AI] really does raise the bar in terms of what people are going to expect from you,” our analyst Jeremy Goldman said on our recent “ChatGPT and Generative AI” panel.

As AI matures, it will become more specialized, automating mundane tasks, ushering in personalization, and changing the way consumers, retailers, and marketers use the internet. Here are eight predictions for that not-so-distant future.

1. AI companies will boom and bust

We’ve entered a “new era of the internet,” according to our analyst Yoram Wurmser, but that doesn’t mean all that glitters will remain gold.

  • AI spending worldwide will increase to $154 billion this year, a jump of over $30 billion YoY. But it will also weed out low-performing companies.
  • “This is the wild west right now in AI companies,” said Wurmser. “If you can show that you have an application for AI, you’re getting money right now. And a lot of these are built on pretty shallow foundations that can be replaced by the next generation of AI.”

2. Models will get more specialized

Right now, a few AI models boast a broad spectrum of capabilities, but in the near future, individual systems will niche down.

  • “I think there’s going to be a lot of emphasis not on building larger and larger models, but building more models that are better aligned with the goals of people that are interacting with them,” said Wurmser. For example, an AI system for marketers would be specifically calibrated to drive, measure, and respond to KPIs such as clicks or conversions.
  • Specialized AI will mark a shift from experimentation into increased efficiency, but that could also usher in job loss.

3. AI will take over for mundane tasks—unless it doesn’t

AI has the capacity to take over tasks like building spreadsheets, drafting small changes for social posts and retail listings, and compiling data.

  • “I think it’s just going to have a massive impact on innovation and productivity,” said Wurmser.
  • But companies may also lean into AI for creative, leaving humans with the grunt work of supporting the AI.
  • A focus on AI assisting people rather than the other way around would be ideal, but it’s not an assured outcome right now.

4. AI will supercharge personalization and product development

AI has the capacity to take customer information and turn it into hyper-specific product recommendations, or even develop products specifically for a given consumer.

AI personalization could present too many options to consumers, resulting in decision fatigue. Or it could creep out customers by predicting their desires.

5. Prompt engineering will inform AI content quality

With current AI models, prompt writing is a need-to-have skill, according to our analyst Dan Van Dyke.

  • Van Dyke advised that content marketers specify frameworks, verb choices, perspectives, and details when working with AI, like asking ChatGPT to write like a marketer with 20 years of experience, or specifying copy should be quippy and light.
  • “Those extra descriptors that you can use to prompt engineer make all the difference when it comes to quality,” he said.

6. Consumers will expect AI disclosures

Consumers already have little trust in Big Tech. “All these businesses have a big task ahead of them to make sure that what they’re doing in generative AI is very plainly disclosed to the consumers that are interacting with them,” said our analyst Debra Aho Williamson.

An influx of AI could result in people expecting that all content is in some way touched by AI, breaking down trust even further.

Or there’s a chance consumers will ignore low-quality AI content altogether. “If there’s more content out there, you’re not even going to necessarily notice as a consumer because you don't have any more time to pay attention to it,” said Goldman.

7. Humans will fill gaps for things AI can’t do

AI will eventually overcome many of its shortcomings, but as it stands, humans still play necessary roles in assuring data privacy, accuracy, and authenticity from AI.

One of ChatGPT’s limitations is privacy.

  • “ChatGPT is very explicit that [it] own[s] all the data that you give it,” said Van Dyke. That means businesses have to be careful about sharing customer information.
  • “I think it’s particularly concerning for retailers in regions with strict data privacy laws, such as the EU, and regulators are already taking a bit of an interest there,” said Perkins.

AI also risks plagiarism, or even being called a “pathological liar.”

  • “Use your oversight to really make sure that you’re not producing content that is misaligned with what you want to do, that the content is not providing you false information, and that the content isn’t plagiarizing something else,” said Wurmser.
  • Van Dyke warned users to watch out for “hallucination,” or total falsehood, and “algorithmic bias” in AI.

In addition, AI can’t provide high-quality customer service for all instances. “The tech also lacks really the kind of emotional intelligence and empathy to deal with situations where the consumer is perhaps getting a bit frustrated or with really nuanced complaints,” said Perkins.

8. Auto-GPT will turn everything on its head

It’s hard to make accurate predictions about AI in marketing and retail, because the field is changing so fast. With that being said, Auto-GPT, where AI functions autonomously without or with little human input, will transform the way people use AI. Auto-GPT could be able to not only come up with a business plan, but deliver on intermediate tasks, or it could create a collection of image-generation prompts and then actually generate those images.

“That far-off use case is maybe not so far off in the future,” says Van Dyke. Experiment with the tech now, but keep a hand on the AI pulse.

This was originally featured in the eMarketer Daily newsletter. For more marketing insights, statistics, and trends, subscribe here.