From The Straits Times    |

If you’re constantly on the go and juggling a million things from professional commitments to personal to-dos, ChatGPT is a godsend. It helps us craft emails, ideate new projects, clean up messy notes, draft interview questions… And we’ve barely scratched the surface. 

The impact and influence of ChatGPT is undisputed. The only platform to have ever reached 1 million subscribers in three days (it took Netflix 3.5 years), it’s changed the way that we search for information and use the internet. 

AI has been pervasive for a while now, from the face ID on our phones to social media algorithms. But how exactly has ChatGPT been a game-changer, and how is it different from existing models? 

Essentially, it is based on a deep-learning architecture called a transformer (FYI, GPT stands for Generative Pre-trained Transformer) which is a type of neural network. It is trained on a large subset of text data, and learns how to predict the next sequence of words. Once it’s been trained, it learns how to generate text using language patterns it has picked up. 

As more data is inputted into the model, the more it improves its performance. It combines generative AI – that is, AI that can create new data – and Large Language Models (LLM) that use natural language processing techniques to generate text. 

To truly understand the wide-ranging impact of AI, and whether it’ll be taking our jobs any time soon, we pick the brains of Jennifer Zhang, founder of Wiz.AI. Founded in 2019, Wiz.AI builds voice AI models to personalise customer service. If you’ve ever gotten a call from a company trying to sell you something in a convincing Asian accent, chances are, you might not have been talking to a human being. 

The launch of ChatGPT has transformed the way we communicate with AI models, as conversations have become more personalised and responses more sentient. For Wiz.AI, this represented an opportunity to further bolster customer service. Jennifer explains, “The difference between TalkGPT with our previous offerings is that TalkGPT is a self-service solution that enables dialogue creation powered by ChatGPT. Where our Talkbot Enterprise solution relies on customer experience (CX) designers to craft dialogues, TalkGPT output comes from generative AI.” 

She goes on to explain that this is a game-changing development that also automatically translates text to other Southeast Asian languages, including, funnily enough, Singlish. 

“It changes the game completely because SMEs can now explore AI solutions the way large enterprises can,” she explains. 

This means, complex tasks can be simplified and executed by AI, thus freeing up time for us to truly focus on what’s important. 

Here, we speak with Jennifer to understand the good and the bad of AI. 

How do you see AI evolving in the next few years, and what impact do you think this will have on businesses and society as a whole?

As I see it, AI technology years ago used to be like a junior student. You needed to train it with specific tasks or workflows. Scaling it involved a lot of work. 

Today, it’s at a “university” level. People can generate content like images, videos, voices, and even codes in a few mouse clicks or finger taps with AI. Generative AI brings a lot of possibilities. This makes AI a lot more accessible. Before it came along, AI was like a black box, complex, difficult to understand and mostly working behind the scenes. Now, anyone can access the technology through their personal devices.

I think ChatGPT and other generative AI tools are helping to open up more possibilities for the average consumer. Currently, a lot of digital transformation is inaccessible to specific demographic groups. But what I see is the next level of AI and tech reaching cross-media, cross-language, and cross-border applications.

With evolution, I believe digitisation will become more accessible. It will level the playing field for smaller companies that couldn’t afford enterprise-level solutions before.

Recently, the Future of Life Institute and Elon Musk have called for a pause on the development of AI beyond GPT 4, citing “risks to humanity”. As there is widespread adoption of ChatGPT, how can users ensure that they use this technology responsibly?

In general, I think users should look at both the possibilities and reliability of new technologies. For example, when autonomous driving was introduced, it got consumers and automakers interested. But the industry became sceptical when it had to consider car performance and road safety. Over time, that has improved, and more companies are looking to make autonomous vehicles more accessible for everyone, not just as a novelty or a luxury.

I foresee the same pattern with ChatGPT. It generated quite a buzz when it was launched, but at the same time, people have pointed out its limitations. Despite its shortcomings, like providing unreliable information, many companies are willing to embrace ChatGPT in their workflows. I think people in general recognize the power of generative AI and they want to see firsthand how ChatGPT can be implemented in their own industries, in specific use cases.

Can you also elaborate more on the possible risks of using AI, and the responsibilities that companies need to bear in mitigating these risks?

Security and data accuracy can be some of the risks in AI adoption, especially for enterprise companies. The data for OpenAI’s large language model (LLM) GPT-4 resides in US servers, and enterprise customers are not allowed to provide data outside their own environment or jurisdiction for security reasons. And when it comes to data accuracy, conversations that are completely reliant on open models cannot fine-tune the messaging or information that’s being provided.

This is why I believe companies or industries will begin to have their own language models to layer on top of the existing models. Having their own language models or area models enables a company’s AI solutions to understand, summarise, and generate content that’s tailored to the industry. It becomes more accurate and because it’s a local model, security will no longer be an issue. 

I think companies that don’t explore and embrace AI will be at risk of becoming obsolete faster than their competitors.

Jennifer Zhang

What advice would you give to businesses that are considering implementing AI in their operations or customer engagement strategies?

AI transformation or digital transformation can be challenging. But I believe that implementing AI as part of a business’ customer engagement strategies can be one of the easy wins in this space. 

Solutions like TalkGPT and Talkbot Lite can be deployed quickly and with relative ease, and results can be seen quickly. The best part of these solutions is it enables you to iterate and fine-tune the messaging to see what kind of tone or word choice resonates with your audience. There are other basic AI solutions that businesses can adopt for their processes and internally at Wiz.A;I, we encourage our employees to explore ChatGPT for their day-to-day work. Really, it’s about using the available technology to boost efficiency and reduce workload. Adapt or be left behind.

There’s also been much talk about  how AI can be replacing jobs – but we said the same thing when calculators were launched. What is your perspective on this, and what can we do to prepare for this shift?

It’s true that AI has the potential to automate certain tasks and processes traditionally done by humans, but I don’t think AI will completely replace humans. Like my earlier analogy of being a student, AI can take on some tasks but it still requires direction from a human being. And as certain tasks become automated, new job opportunities will emerge that require different skills and expertise. 

What will separate the industry leaders from the followers is the ability to leverage AI in their workflows and processes. Exploring the capabilities of AI is essential for companies that want to remain competitive and stay ahead of the curve. I think companies that don’t explore and embrace AI will be at risk of becoming obsolete faster than their competitors.

We talk about the risks – but AI has also done some good. Can you share some examples?

One of the areas where AI has been used for good is in healthcare or in medical research. For example, developing cancer medicine is long and expensive due to the complex nature of drug development, the high standards required for drug approval, and the significant investment required in research, clinical trials, regulatory approval, intellectual property protection, and manufacturing. Drug formulation itself – the combination of various ingredients in the right proportions – is difficult to get right. 

People in pharmacology understand that interactions between different elements can determine the effectiveness of a drug, and just one small change can result in a massive difference. What’s more, multiple combinations need to be studied, modelled, and tested to arrive at an effective drug. 

Doing all this the old way would have taken years. With generative AI, the research and modelling can be done much faster, and drug developers can move forward into testing much sooner than they otherwise would. This enables them to reach medical breakthroughs at a faster pace. That’s just one area where AI is bringing massive benefits to humans.

What excites you the most about the future of AI and its potential to transform the way we live and work?

AI really has the power to disrupt how we do things. Take for example the Business Processing Outsource (BPO) industry, which is traditionally reliant on human labour. Today, BPO companies are rapidly developing and adopting AI in their processes to improve productivity and accuracy. They are realising that AI technology is not an enemy to their human agents but an active helper, freeing up employees to focus on more complex and valuable tasks.

AI can also help to identify and resolve errors in BPO processes by analysing vast amounts of data and identifying trends that can help streamline future operations. Overall, the possibilities in AI are endless. And I think generative AI will play a large part in transforming a wide range of industries and applications, including customer engagement.