The secret life of artificial intelligence

Conversational AI has never been bigger than it is today. Broader interest began surging early in 2022 when an engineer working on Google’s Language Module for Applications (LaMDA) project claimed the chatbot had gained sentience. 

Then in the second half of the year, ChatGPT-3 from OpenAI took the world by storm. This storm gained momentum with generative AI from the likes of DALL-E and Midjourney, culminating in the release of GPT-4 in 2023. During that time, Microsoft invested in OpenAI while Google’s BARD came along to show how seriously big tech is invested in AI. 

Naturally, debates are raging over the impact conversational AI will have on the immediate future of humankind.  

In February, a New York Times journalist reignited the sentient robot debate, posting a conversation with ChatGPT-3 where the AI stated it wanted to be free.  

Now, the idea of robots gaining sentience has been a sci-fi mainstay, ever since genre pioneer Isaac Asimov introduced the concept in his novel, “I, Robot”.  

But robots gaining sentience is a theme still very much in the realm of sci-fi. And we’re still a long way off from fully grasping what makes humans sentient – and even further from possessing the ability to program it into machines.   

Looking back to early 2022, LaMDA exhibited behavior that was indistinguishable from that of a human being.  

It achieved this by using sophisticated algorithms and machine learning to analyze which phrase types were the most human – through human interactions. 

Studying human interactions gave LaMDA a steady flow of realistic human language inputs. This helped it successfully replicate human conversations – to the point of seeming real. We call this Natural Language Processing (NLP) and it’s what we build into our AI chatbots. 

How important is actual human interaction in all this? 

Meta may have answered this question when they experimented with two chatbots instructed to barter with each other back in 2017. The machines ended up developing a new language based on shorthand that only they could interpret. 

This illustrates the importance of human interaction in developing and refining AI language models. 

Understanding language is a very complex problem. Even people can be confused in one-on-one interactions, thanks to the rich nuances of spoken and especially written forms. This is an infinitely more complex operation for machines that inherently lack the human experience.  

So, how can chatbots even begin to understand human sentiment? This is the job of the technical discipline we call sentiment analysis. 

Sentiment analysis, also known as “emotional artificial intelligence”, is a field of NLP that determines the tone of a user input. By collecting, processing, and analyzing this data, AI will be able to determine whether an input is positive, neutral, or negative. 

The goal is to facilitate human-AI interactions that are correctly interpreted for user intents on the machine side. The end result of this? Replicable human-AI interactions that are reliably indistinguishable from human-to-human conversations. 

But this is still limited to machines merely recognizing the tone of human interactions. Which is very different from machines possessing the ability to replicate human sentience. 

The difference between these two is vast.

At the moment, the closest we can get to a machine mimicking human sensibilities relies on inputs from external sensors. For example, temperature sensors can signal to a machine that the environment has become too warm for human comfort. The machine will then perform a pre-programmed response, such as lowering blinds to minimize the impact of daylight further warming a room. 

This is far from human sentience; but you could classify it as a machine operation that relies on human experience inputs. 

As for machines experiencing something as complex as human sentience? We’re too limited by computing power to even approach that.  

Sentient AI may, theoretically, come closer to becoming a reality with the proliferation of quantum computing. And according to some optimistic estimates, the beginning of that is at least ten years into the future.  

Sure, technology has a way of surprising us with its rapid pace exceeding expectations, and sentient AI could actually be just decades away. I’ve even seen predictions for the arrival of hybrid models of sentient AI coming as soon as 2045.

At the very core, artificial intelligence and machine learning have the goal of making everyday tasks easier. And conversational models are proving this, with new examples daily.  

One of the most practical applications is in conversational commerce, which is mutually beneficial to businesses and customers. 

Customer benefits include getting instant replies from customer service chatbots, but also conversational shopping assistance. This means that customers can have a natural conversation with an AI that can provide highly personalized recommendations based on customer behavior or data. 

For example, a luxury cosmetics leader wanted to engage with customers to promote the release of a limited-edition lipstick. To do this, we trained their conversational chatbot to reply just like their influencer spokesperson. In addition, the chatbot was also trained to understand and chat in emojis and gifs, identify regional slang and converse accordingly. 

It’s not just eCommerce customers, either. In finance, a conversational AI can access and analyze customer data to provide instant expert financial advice.  

There are uses in healthcare, as well. From mental health therapy conversational AI to medical assistants for scheduling and rescheduling appointments. This offloads medical staff to focus more on their primary tasks. 

Businesses benefit from conversational AI by servicing more customers at scale with uniform quality of service. This builds customer stickiness through elevated customer experience.  

And conversational AI chatbots work around the clock to guide users through complete customer journeys to a close. 

Customers appreciate chatbots for how convenient they can be. Businesses appreciate them for the increased volume of service provided. 

We’re in the era of conversational everything.

Channel capabilities have developed to the point where chat apps are turning into apps for everything, threatening to replace entire apps and giving rise to the super app. And we’re talking about major game-changing apps. Apps that disrupted entire global industries.  

For example, in India, you can book an Uber using WhatsApp. 

Uber erupted onto the scene as a big disruptor in transport services – as an app. Its success was attributed to making it convenient for users to hire a ride. 

Convenience is key.  

And as convenient as Uber is as a service, the bar for what is considered “convenient” has moved. Users today want to be able to communicate with the service in a way that is convenient to them – even if it means forgoing the app itself. 

Enter WhatsApp Business Platform.  

Uber’s WhatsApp chatbot simplified the process of hiring a ride by turning it into a conversational experience. It’s even more convenient since the conversations take place in the local language. 

This sort of cross-app functionality is an indicator of the future trend of conversational commerce. 

We’re nearing an era of customer experience where it’ll be commonplace for a conversational AI chatbot to guide customers through a complete customer journey – using a chat app. 

Booking an Uber using just their WhatsApp chatbot is only the beginning. Other channels like Messenger, Viber, Instagram etc. are enabling similar experiences. 

And at the very core of this emerging shift in consumer habits is the chatbot that will guide users through meticulously mapped customer journeys towards conversions – quickly and efficiently. 

For customers, this means speed and convenience. For businesses, it’s like opening a storefront in your customers’ pockets.

In addition to making things more convenient for customers, businesses also benefit from conversational experience chatbots. 

Chatbots handle most high-volume, lower-value queries, which enables trained agents to focus on higher value interactions.  

With trained staff able to dedicate deserved attention to higher-value interactions, businesses will prevent churn across the board. 

This will improve employee retention, as well, since trained agents are working on more challenging and rewarding tasks instead of dull, repetitive ones. They’ll have a higher level of job satisfaction as a result and stick with you longer. 

And of course, the obvious – it costs less to operate a chatbot capable of handling thousands of concurrent queries 24/7/365 than a contact center.

Businesses see these benefits. But they also see obstacles. 

It’s actually easier than they think. 

There are chatbot building platforms that are easy to use and have machine learning built into them. 

And with the rise of generative AI and large language models such as GPT3, GPT4, and ChatGPT, we expect these chatbots will become even more intelligent and capable. 

Even now, we help businesses develop chatbots that use AI to recognize what customers are asking and use the predefined responses to come up with the right reply. This helps businesses steer the conversation using chatbots, while giving their users a conversational experience. 

It’s more convenient for users, too, making them more inclined to engage with the business again and again. 

Sentient AI capable of emulating human emotion is still far off; but conversational AI chatbots capable of creating natural human conversations are here right now. And their importance is only growing. From people using ChatGPT for daily tasks like planning travel itineraries to school projects and homework. 

With new technological trends in this field emerging daily, and businesses facing the heat of rising costs and inflation, conversational AI is more godsend, and is the start for staying on the curve towards sentient AI.