ChatGPT has also brought innovation to conversational marketing technology solutions as a large language model (GPT) that learns to steer conversations and generate text similar to human speech. In the past, we used to rely on a tree structure for smart conversational bots, while today the process of programming and integrating a bot is a little different. However, we are still dealing with a bot that needs to be taught something. It is, after all, a generative artificial intelligence that can speak in a human-like manner, form sentences into a coherent whole and convey meaning. We sometimes joke that ChatGPT hallucinates. It's true – if you don't train your bot properly, give it all the information, connect it to your knowledge bases and, if necessary, provide it with the support of a live agent who becomes its AI trainer, it will start making things up. While that’s great news if I want to write a bedtime story for my five-year-old daughter, it is certainly not the approach suitable for serious companies and influential brands.

I firmly believe that conversational AI is unlocking a world of endless possibilities for brands, customer service and conversational commerce, rapidly transforming many industries with its user base of over 100 million. But the technology solution needs to be properly programmed and trained. In practice, this means that a company inputs its data into a (closed) knowledge base, a so-called 'prompt engineer' processes it and trains the bot accordingly. This process involves posing the most frequently asked questions to the bot and checking that the answers are correct, adding new information, monitoring the most frequently asked questions and answers, and populating the knowledge base. 



In any case, I am an advocate of private data models driving such technological solutions. When it comes to the development of autonomous vehicles, for example, it is important that the developers use open databases that are dynamic and contain anonymised information that is added to the system on the basis of a higher interest. Let’s look at Google Maps. As a popular AI-driven tool, it mostly provides correct and highly accurate information on road conditions and recommended detours, which is why it has served as our faithful assistant for many years.

I am convinced that the advent of new AI-driven solutions will revolutionize the way we consume information and the way we organise our work. As AI technology becomes more widely used, it will first of all noticeably change the way we search for and discover information, and the way we consume media online. In the future, websites will be partially replaced by smart chatbots or virtual assistants, which will solve customer challenges by talking to them, and of course in a completely autonomous way. Just consider the capabilities of Siri today; not only does it quickly translate between languages, it can also recommend you where to have lunch nearby and even answer questions that have a psychological, interpersonal component. At work, ChatGPT can already be integrated into various other applications via an API to organise your schedule, take meeting notes, edit documents and more. The immense potential of AI and the remarkable processing power of AI simply cannot be overlooked.

The column was originally published in Marketing Magazine, Nr. 505-506, July 2023.