almohannad.cx
Customer Experience

Deploying AI in Customer Experience

8 min read Translated from the Arabic original

Introduction

On the sidelines of my participation as a speaker in a panel discussion at the CX World Forum, I wanted to share an article about the topic so the benefit can spread and those who missed the chance to attend the conference can benefit.

There is no doubt that AI is a revolution in every aspect of our lives, and we have to grasp both the benefits that come from using its tools and the risks they bring. We need to know well how, through these tools, we can multiply our productivity, through awareness of the available tools (free and paid) and mastery of the prompt engineering associated with these tools.

Human Interaction

No matter how far technology takes us, interaction with people will always have a special flavour that no technology can replace. Human communication will remain the highest form of communication and the most authentic, and we, as humans, will remain able to distinguish texts, responses, or even voices generated by an AI tool, at least for the next five years.

In the midst of the AI sweeping through the trends, and in the midst of leaders’ and decision-makers’ obsession with automation and digital transformation, human interaction will, over time, become a competitive advantage that a small number of companies will boast about providing as added value on top of their services.

The Ethical Side

There are good practices (I won’t call them “right”) and bad practices (I won’t call them “wrong”). Relativity applies here in an objective way that varies from one person to another, or from one company to another, based on their value system and the ideals they believe in.

It was a scene to laugh at and cry at, that university graduate who sent a thank-you message to ChatGPT.

I wonder about the level of students who will rely throughout their university studies on generative AI tools. How will they fare in the absence of an internet connection or without access to one of these tools? Will their abilities crumble at the first problem they face?

Generative AI tools still fall into the trap of hallucinations, meaning they invent information and invent sources and names that don’t exist. It’s a matter of time, and these tools will overcome this issue, but using them without disclosing that has an ethical dimension.

All AI systems are based on open-source data, and their engines are trained on certain inputs to produce certain outputs. What is this data? Are its rights protected? What are its sources? How accurate is this data?

Unintended bias and discrimination because of algorithms: if AI systems are carrying out certain actions, have they been tested and confirmed not to make any biased decisions when they pass judgements? Have the appropriate adjustments been made to confirm that no such bias exists if it was found during testing?

The Saudi Data and Artificial Intelligence Authority (SDAIA) has published two documents to govern the use of these tools as much as possible:

Good Practices

  • Educate users on the principles of these tools, their risks, and their weaknesses, in addition to teaching them prompt engineering.
  • Educate users on the kind of data that can be fed into generative AI tools and the kind that cannot be used with these tools because of confidentiality.
  • Automate certain procedures (greater accuracy, higher speed): for example, processing financing applications that require calculating the applicant’s financial risk, account-opening procedures, and fraud-suspicion pattern detection. None of these require high mental capacity, yet they used to take a great deal of time to process.
  • Sentiment analysis from social networks: instead of having a person or a team sit down, or even instead of using prohibitively expensive tools to categorise and classify a set of tweets, AI tools can do this job in a matter of seconds.
  • Analysing customer care agents’ calls (after converting them from speech to text) and guiding the agent with real-time tips and prompts based on their performance in that call.
  • Partial reliance on generative AI tools to generate first drafts of micro-copy that will be used as starting points for UX writing.
  • Partial reliance on these tools to rephrase drafts prepared in advance by humans, with the aim of altering their tone and voice to be more aligned with the brand personality.

Bad Practices

  • Relying entirely on generative AI tools without any human intervention, specifically in content generation (I mean copy-pasting straight from the tool), which is unethical. LinkedIn co-founder Reid Hoffman wrote a book using AI tools, but at least he disclosed that. Many other people publish articles, posts, and books written entirely using AI tools without giving the slightest hint that they relied on them.
  • Chatbots that rely entirely on generative AI tools and play a role in delivering misleading information, even if it looks logical in its structure and form. The right move here is for these tools to rely on closed-source data (such as the database and information of the company providing the service itself).

The Future of Generative AI Tools

AI is expected to make significant leaps in areas of customer experience that were previously challenging.

In 2024–2025, the focus in AI for customer experience (CX) will shift from being merely a novel idea to a useful tool that must be used. Last year introduced the public to AI capabilities such as ChatGPT, making many of its tool names familiar.

This will include AI improving its understanding and anticipation of customer needs more accurately, providing personalised experiences, and automating more complex customer interactions. The story will be about AI’s evolution from a cool concept to a vital tool in enhancing customer engagement and satisfaction.

Personal use case 1: I used ChatGPT-4 to identify the most suitable candidate for a particular job. I uploaded the job description for a role, then uploaded the candidates’ files with their biometric assessment results, such as a personality test and the Clifton StrengthsFinder test. The tool gave me a striking analysis of each person’s strengths and weaknesses, who was most suitable for the role and why. This would have taken me at least a full day of work (a task I completed through ChatGPT in less than 10 minutes).

Personal use case 2: I used ChatGPT-4 to generate micro-copy drafts for UX writing after writing a long, tedious prompt that included our brand personality and tone of voice guidelines, along with a request for multiple options tied to influencing the message recipient in different ways. I ended up, in a matter of seconds, with several drafts, from which I created one outstanding final version in a record time compared to what I used to spend doing it manually.

How Can Companies Get Started With Generative AI Tools

The real opportunity lies in unlocking the full potential of AI by understanding its capabilities and using a variety of tools for different use cases, and embedding these tools into business processes.

  • Start with intention: the solutions offered through AI tools don’t create value for beneficiaries unless their needs are placed at the centre. Our intention when we want to use AI tools should be to add meaning and value to the lives of customers or employees. Lowering costs will be a by-product, and there’s no value in lowering costs if it will negatively impact the human experience.
  • Understand the capabilities of AI tools: it is essential to get to know the strengths and limitations of current AI technologies. This understanding allows companies to match the right AI tools to the right tasks, enhancing operational efficiency.
  • Identify use cases for AI tools: every AI tool has specific functions that make it suitable for certain tasks. Companies should consider using a variety of AI tools in their workflows, selecting each based on its suitability for a particular task. Multiple tools can also be used sequentially or simultaneously to perform a single task.
  • Embed AI tools into business processes: integrate AI tools into business process documentation. This integration helps create a streamlined workflow where AI tools complement manual processes but do not replace them.
  • Plan for contingencies: it is essential to include manual processes in procedures alongside AI tools as a contingency for scenarios where there is no access to the internet or to AI tools, ensuring work continues without interruption.

In Closing

In the end, it’s worth emphasising that generative AI, with the possibilities and challenges it carries, represents a revolution that cannot be overlooked in the world of customer experience and business in general. The use of these tools should be aimed at improving customer experiences and increasing operational efficiency, while always preserving human and ethical considerations. Striking a balance between technical innovation and human values is the key to making the most of these technologies.

It is essential we always remember that technical tools, no matter how advanced, are fundamentally just tools in human hands. Their effectiveness and value are determined by how we use them. So we have to recognise the importance of the role we play as users and developers in this rapidly evolving world, where ethics and human interaction remain at the heart of everything we do.

We have to invest in educating and training individuals in prompt engineering and in understanding the fundamentals of AI, with an emphasis on the importance of preserving human interaction and ethical integrity. By doing so, we ensure that we’re heading toward a future that embraces technological progress without sacrificing the human essence that distinguishes us.

Disclosure: I rephrased parts of this article, and also designed the article’s cover image, using ChatGPT-4.

← Articles MOC