How Business Leaders Can Leverage Generative AI in Customer and Employee Experience

0
504

Share on LinkedIn

The AI playground has evolved a lot throughout the past year. Conversational AI has made huge inroads. We have seen an increasing number of deployments and through their availability, people get to use them more and more, such as Apple’s Siri, Amazon’s Alexa, or Google’s Assistant.

I see it in my own family; we are increasingly asking Siri for something instead of opening a browser and typing our search query.

At the same time, the little windows of chatbots have become quite ubiquitous on websites.

And then, the next iteration of AI hype had arrived: Generative AI. We were introduced to systems like Dall-E or chatGPT of Open AI or You.com that also offers search, to name only a few of the best-known ones. In the meanwhile, Microsoft has also integrated chatGPT into Bing, while Google has suffered from the probably most expensive demo ever when its generative AI gave a wrong answer. Apps like Lensa popped up after ones like SnapChat offered AI-based filters for some time.

Of course, I ventured and tried services, generated images for some of my blog posts, got into some “discussion” about web3, even had it generate a sample terms and conditions document for a consulting agency. Guess what, it came with a lot of disclaimers.

AP is using generative AI for some years now while CNN seems to have overdone it in a recent experiment. Google Duplex got deprioritised some four years ago as it was a bit too good, i.e. creepy.

Some people already ventured to say that generative AI will supersede conversational AI. Will this really be the case, or is this only the next craze?

This leads to an obvious initial question.

What is a conversational AI vs. a generative AI?

I mean, apart from conversational AI not being able to create pictures.

Who better to ask than ChatGPT, You.com, and Google? This also gives a good chance to see how easy it is to get an answer to the question: “What is the difference between generative AI and conversational AI?”

Google, of course, provides several well-fitting links that I can use to figure it out myself. No surprise here.

ChatGPT replied as follows:

“Generative AI and conversational AI are both subfields of artificial intelligence, but they have different applications and focus on different tasks.
 
Generative AI refers to the branch of AI that focuses on creating new content or data. This can include creating images, music, text, and more. These models are trained on large datasets and use statistical patterns to generate new, unique content that is similar to the training data. Examples of generative AI include GPT-3 (which I am based on) and other language models, and Generative Adversarial Networks (GANs) used in image and video generation.
 
Conversational AI, on the other hand, is focused on creating systems that can understand and respond to natural language input from users. These systems can include chatbots, voice assistants, and other interfaces that use natural language processing (NLP) and machine learning to understand and respond to user requests. Examples of conversational AI include Apple’s Siri, Amazon’s Alexa, and Google Assistant.
 
In summary, Generative AI generates new and unique content while conversational AI focuses on understanding and responding to natural language input from users.”

The ‘You Write’ service of You.com made the following blog style statement for professionals in a neutral tone.

AI technology has come a long way in the past few years, and with it have come two distinct types of AI: generative AI and conversational AI. While both are powerful tools that can be used to automate tasks or provide intelligent services, they do so in different ways.
 
Generative AI is focused on creating new content from existing data sets. It uses algorithms to generate patterns based on what it already knows about a given topic or dataset. This type of artificial intelligence can be used for things like natural language processing (NLP) applications such as summarization, translation, question-answering systems, and more. Generative models are also commonly used for image recognition tasks such as facial recognition software and self-driving cars which use deep learning techniques to recognize objects around them in real-time scenarios..
 
Conversational AI focuses more on understanding user intent through natural language interaction by using speech-to-text technologies like voice assistants (e.g., Alexa), chatbots, or virtual agents that interact with customers over text messages via SMS or messaging apps like Facebook Messenger, etc. Conversational AIs use NLP techniques similar to those found in generative models but instead focus solely on responding accurately based on input given by users rather than generating new content from scratch. These bots usually rely heavily upon predictive analytics methods such as machine learning algorithms which allow them to learn how best to respond when presented with certain inputs/questions over time. Generally speaking, this type of artificial intelligence is designed specifically for customer service purposes where accurate responses need quickly provided within short periods without any human intervention necessary.
 
In conclusion, while both Generative AIs and Conversational AIs share similarities due to their reliance upon Natural Language Processing techniques, they each serve a unique purpose depending upon the task at hand; one being able to create completely original pieces of work while other providing quick automated responses within specific contexts respectively.”

Asked for a paragraph in a neutral tone for a professional audience You.com gave a shorter summary.

“Generative AI and Conversational AI are two distinct types of Artificial Intelligence (AI). Generative AI is focused on creating new content, such as images or text, while conversational AI focuses on understanding natural language. Generative models create data that can be used to train other machine learning models, while conversational models use speech recognition technology to interpret human conversations. To write well with either type of Artificial Intelligence requires an understanding of the strengths and weaknesses associated with each model. Additionally, it is important to consider how best to communicate your message in a way that resonates with both humans and machines alike.”

In summary, both use machine learning, in particular deep learning. Both need some ability of natural language processing. Conversational AI has text/speech output whereas generative AI can also create images or music. Conversational AI focuses on understanding user intent and to react on it, generative ai is about creating an output based on what it was trained on.

And then there are some major differences: Generative AI is using some “world knowledge” that is provided by large language models to generate results, but it is not connected to any business system. A conversational AI is; it is also an orchestration layer. Where generative AI creates human-like responses, conventional chatbots as delivered by current conversational AI are somewhat awkward without putting very high effort into them. Generative AI also sometimes tends to create some fiction — with utter confidence. This is to quite an extent the result of them not being task-bound like conversational Ais. They do not know about any business, which means that their knowledge needs to be augmented by business-specific knowledge.

The generation of output, answers, is where the overlap is. In the business world there is always a context. Users want a response to an inquiry. The information needed to give this response is usually buried in a heap of data.

However, generative use cases are broader. In a business environment, conversational AI so far tends to focus more on service and self-service type of scenarios, whereas, in theory, generative AI can support a whole variety of other scenarios, too, including creating marketing materials and, hey, whole blog posts as part of it. Of course, there are still limits, as the examples above also show.

So, in a sense, generative AI augments conversational AI but does not replace it. The two technologies form a Yin and Yang type of solution.

How to leverage generative AI in a business?

The obvious first stage is the improvement of customer service as well as external and internal self-service scenarios. Where the conversational AI takes care of the structuring of the customer request, the connection to the various business systems, and the search, the generative AI builds the answer in a more human form, extracting the relevant info from the search results and putting them into a concise answer. This is true for customer self-service as well as for assisted service, where the agent gets the support of the AI.

Another one is user training. There are usually treasure troves of information buried in wikis, chats, and documentation systems, even file systems. This can be used by answering many user questions, therefore tremendously increasing productivity and reducing friction.

A bit further down the line lies the generation of standard responses to emails and the creation of marketing content, first in short form, then in longer form.

I am also keen to see new types of no-code and low-code solutions that work with spoken input to create working solutions. This not includes makers of conversational AI platforms.

What should leaders do?

Considering that customer experience and employee experience are the biggest levers for continued business success, the easiest wins are in enterprise search, customer self-service, and agent support. The customer self-service scenarios can very well lie in the lead-generation area, too. In each scenario, work with your vendors on how their solutions can help offer concise answers to questions instead of only pointing to documents. Leading vendors like Cognigy surely have a good answer.

Formulate KPIs that you want to improve and build a case around them. One of the KPIs needs to be the accuracy of the responses given by the system. There is no point in causing rework.

After this, identify the most pressing scenarios and build a prototype. Measure the outcome and take it from there, one step after the other.

A bit more enterprising? Try personalized marketing messages that are created by generative AI, in combination with generated content like easier blogs and product descriptions. Again, do this in a controlled environment, defining and measuring KPIs that you want to improve.

The technology emerged and will be there to stay. Use it with a think big — act small mindset and regularly re-prioritize.

Thomas Wieberneit

Thomas helps organisations of different industries and sizes to unlock their potential through digital transformation initiatives using a Think Big - Act Small approach. He is a long standing CRM practitioner, covering sales, marketing, service, collaboration, customer engagement and -experience. Coming from the technology side Thomas has the ability to translate business needs into technology solutions that add value. In his successful leadership positions and consulting engagements he has initiated, designed and implemented transformational change and delivered mission critical systems.

ADD YOUR COMMENT

Please use comments to add value to the discussion. We will not publish brief comments like "good post" or comments that mainly promote links. All comments are reviewed by moderator before publication.

Please enter your comment!
Please enter your name here