Feature
The Gist
Platform diversification. Don't rely on just one AI platform; exploring multiple platforms can yield better prompts.
Prompt engineering. Understanding various types of prompts and their nuances helps marketers showcase the value of AI-integrated products effectively.
Customer feedback. Assessing customer perceptions of AI-based experiences and addressing concerns can improve overall customer sentiment.
With all the rapidly emerging buzz around ChatGPT4, Bard and other AI-enhanced solutions, it's natural to assume that marketers may feel like they're back in school, constantly cramming for an exam to learn how to best use these tools.
However, smart students know that the most valuable learning happens outside of the classroom. For today's marketers, the key lies in dedicating extra time to experimenting with prompts on various AI platforms. No single platform can provide prompt responses that address all possible factors affecting the customer experience.
Your best prompts will not come from just Bard, ChatGPT, or any one platform.
Observing how prompts work across platforms is the best educational tactic marketers can adopt this year, in order to develop smarter customer experience strategies for the future.
Related Article: Prompt Engineering and Its Vital Role in AI-Driven Marketing
Exploring Our Surroundings to Identify a Starting Point
Been Around Awhile
It might appear that the concept of prompts is exclusive to newly introduced tools — such as Adobe's Firefly or Alibaba's ChatGPT-like service. However, the world has been experiencing AI-like capabilities for quite some time. Numerous devices and software currently on the market have incorporated prompt-like features; it's just that people haven't always recognized them as such.
Think Auto Entertainment Systems
Consider automobiles as an example. In recent years, luxury vehicle brands have incorporated interactive entertainment systems that utilize prompts to operate their interfaces. This technology has evolved into a branding feature to improve customer experience and distinguish brands from one another. BMW, Mercedes and Lexus all have a unique key phrase that drivers use to activate the system, similar to Apple's "Hey Siri" or Amazon's "Alexa." The driver says "Hey Lexus" or "Hey Mercedes" as a prompt. Once activated, the system remains ready to respond to the driver's requests, such as finding nearby service locations.
Stitching It Together
The interface uses AI to stitch together the verbal requests it hears, treating them as prompts, and selects from a wide range of information - from smartphone interactions to location listings — for its response. This ability to integrate information stems from the algorithm's capacity to evaluate tokens, which are small groups of characters representing a word or a part of a phrase. The underlying AI programs learn the position of each token within the prompt and determine how those tokens correspond to its trained data model.
Looking at Context
The prompts in automotive entertainment systems illustrate how people are using large language models (LLMs) in conjunction with nearby devices and contextual factors. These interfaces highlight the importance of considering creative verification methods that are user-friendly, accounting for environmental influences and identifying inappropriate situations where prompts should not provide guidance.
Hallucinations & Co.
For instance, LLMs can experience "hallucinations" — suggestions that seem reasonable but are impractical in practice. Large language models generate information from text without being aware of their own knowledge gaps. They may lack deductive reasoning in some cases, even when the user provides examples or additional prompts. An automotive interface responding to a prompt for nearby restaurants might suggest a permanently closed establishment or an entirely different location, such as a charging station. The way the model addresses errors can significantly impact the customer experience with the AI feature being introduced.
Learning Opportunities
Webinar
Jun
6
The Future of Social Media: Emerging Trends and How to Stay Ahead of the Curve
The impact of emerging technologies like AI on social media
Webinar
Jun
7
The Building Blocks of Personalization
This webinar will explore how personalization can create powerful ecommerce shopping experience.
Webinar
Jun
8
Attracting—and Keeping—Your High-Net-Worth Investors
Join PwC to dissect their recent survey findings on a specific customer subset — high-net-worth investors.
Webinar
Jun
13
The Future of Contact Centers with AI at the Helm
Learn how to balance human intuition and interaction with the potential uses of AI to create delightful and lasting CX
Webinar
Jun
20
3 Steps to Unlock Your Customer Data and Drive Revenue
Discover how Twilio Segment’s #1 Customer Data Platform helps implement data-driven strategies.
Webinar
On demand
AI for Marketers: Optimize Personalization Across the Funnel
Understand the benefits of utilizing AI across your entire marketing stack
Related Article: Top 5 ChatGPT Prompts for Customer Experience Professionals
What the AI Experience Should Be Delivering to the Customer Experience
Customer Perceptions
The diverse range of applications is the reason brand managers need to consider how customers perceive information presented through AI-based experiences when evaluating such products. To do this effectively, marketers should familiarize themselves with prompt engineering and understand the nuances that emerge from using various prompts. By gaining a deeper understanding of different types of prompts, marketers can better showcase the value and benefits that a brand's AI-integrated product offers to customers.
Refining Prompts
Each platform offers differing heuristic options to refine a prompt. Heuristics are steps proceeding to a solution through trial and error. One example, MidJourney prompts, was introduced during a special Zoom event titled /prompt.camp, hosted by Hajj Flemings of Rebrand Cities and Vincent Hunt of The Bureau of Creative Intelligence. Throughout the presentation, Flemings and Hunt explained various aspects of prompt usage and key concepts in prompt engineering.
MidJourney Prompts
MidJourney prompts consist of four elements: content type, description, style and composition. Content type pertains to media-rendering phrases like "high definition." The description refers to the subject, along with its attributes and surrounding environment, such as "a black and yellow Air Jordan shoe worn by a point guard on an NBA basketball court." Style encompasses artistic details that address the cinematic treatment of the description, including lighting. Composition delves into camera specifics, such as aspect ratio, camera view, and resolution, with phrases like "Ultra Wide Angle" serving as potential prompt details.
Interface Design & More
Interface design can also impact prompt selection. To access prompts in the MidJourney Discord account, users must type "/imagine" and then describe the image using the appropriate prompt structure. MidJourney subsequently generates four different images, allowing users to choose the one they would like to refine or upscale further.
Comparing AI Experiences
As more AI tools for diverse purposes emerge, opportunities arise to compare AI experiences and identify ways to streamline workflows through prompt selections. Marketers should assess how different AI platforms perform prompt engineering on content. This examination highlights available optimization options and can bookend decisions for how AI is used.
Gathering Feedback
When presenting AI-based features to customers, marketers should examine how customers perceive their experiences. For instance, did the AI application contribute to a positive shift in customer sentiment? The feedback gathered can help address the most frequently mentioned customer concerns and inspire ideas on how to respond effectively and appropriately.
Avoid This
One crucial lesson for marketers is to avoid using personally identifiable data or proprietary programming. Failure to do so can result in serious privacy breaches or loss of vital assets, as exemplified by the incident at Samsung where programming code was accidentally uploaded to ChatGPT online. Most open-source AI platforms offer options for setting up sandboxed access to the underlying code, so marketers should explore these technical options first. Alternatively, and for a simpler approach, they can create dummy samples of any media they want an AI platform to examine.
Final Thoughts on Exploring AI Prompts
All in all, now is the time for marketers to experiment and discover the potential of AI. They must approach this technological challenge with diligence so they graduate to better customer experience strategies successfully.
