The future of chatbots: the race between ChatGPT and Google Bard

It was only just the size of a small beach ball, but when Sputnik-1 became the world’s first artificial satellite it marked the start of a new era of competition in the field of technology. 

Back in 1957 the Soviet Union had fired the starting pistol on what would become a space race with the United States.

Now more than six decades later a new technology race has started, a global competition to perfect a chatbot artificial intelligence (AI), and this time it is tech companies – not nation states – which are leading the way. 

What is an AI chatbot?

Chatbots in their most basic form have been around for years and are essentially just a piece of software able to respond to questions. But up until now, those questions had to be pretty simple, think something along the lines of ‘how much is my phone bill this month’ or ‘what is the capital of Australia’. But advances in artificial intelligence have given rise to a new generation of AI-powered chatbots that can write articles (not this one, by the way), compose music, answer complex exam questions and even code. Experts think these could just be the tip of the iceberg with AI chatbots tipped to disrupt a wide range of industries. 

But its real worth could be in the field of Internet search. Such is its ability to sift through the flotsam to produce super accurate answers, it has already been dubbed by some as the ‘Google killer’.

The main challengers

Just as the USSR and USA led the way in the space race, the chatbot AI race is being led by two heavyweight competitors. The first to hit the ground was ChatGPT, developed by San Francisco based company OpenAI and backed by Microsoft.

What makes it different from what’s come before is that ChatGPT uses pre-trained generative chat that harnesses natural language processing (NLP). This enables it to respond to written queries in the style of a human, using sources of information like text books, websites and articles to provide answers. Microsoft said it wants to use it to enhance its Bing search engine and Edge web browser. In February 2023, the tech giant also announced it would release technology for big companies to create and customise their own chatbots using ChatGPT.

Google’s answer to ChatGPT is called Bard. It harnesses Google’s own language learning model Lambda. You might remember it from the news recently when a Google engineer claimed Lambda was so intelligent it was now actually self-aware (similar to Skynet in the Terminator franchise, though hopefully with less drastic consequences). Although this claim was refuted by Google and the engineer later fired.

The biggest difference between the two is the sources of data they use. ChatGPT only uses sources which end in 2021, but Bard is able to access information from the Internet in real-time.
Google views Bard more as a tool to augment its own search function, but it is expected to be utilised by businesses to provide automated support and human-like interaction.

Others enter the race

But the two big hitters are by no means the only creators of AI chatbots on the scene, with Pandora’s box now very much open. In addition to Bard, Google has also announced a $300m investment in a company called Anthropic, which is developing its own chatbot.

Facebook’s parent company Meta has also launched its own chatbot called Blenderbot. There are also several startups companies working on chatbot AIs, including Jasper AI, ChatSonic, Wordtune and OpenAssistant. 
And it is not just US companies which are developing the technology either. Chinese tech giant Baidu said it is developing a chatbot called Ernie Bot - also known as Wenxin Yiyan - which it plans to roll out in March).

Is the buzz justified?

So after all this should writers, lyricists and college professors despair? Perhaps not quite yet. While the technology has shown undoubted early promise it is far from foolproof and still limited.

It is so far text only and ChatGPT is restricted to using Internet content from 2021, which does not update. Chat AIs are also not able to fact check, using sources and presenting the final published work as fact.

Some experts claim that because of the very nature of the large language model, which are fed datasets composed of billions of words which are used to train AI – similar to a super complex version of predictive text – it will still be prone to errors and require some kind of human supervision and editing.

Indeed, some of the tech’s limitations were rather publicly exposed earlier this month when Google provided a demonstration of Bard. It made a mistake answering a question aboutt he James Webb Space Telescope’s recent discoveries, which drew comparisons with some of ChatGPT’s limitations, and the price of Google’s stock fell several points as a result.

Availability of the tech

ChatGPT is available in free and paid-for versions but Google Bard is currently only available to limited beta testers but not yet to the wider public.

What does the future look like?

Much like the launch of Sputnik-1, the rollout of ChatGPT feels genuinely like a milestone. An AI of its capabilities could revolutionise Internet search, academia, programming, journalism and marketing to name just a handful of areas.

That being said, the Hollywood view of AI as something which is able to ‘think’ for itself is still some way off, and even some of the most pro AI computer scientists doubt it will ever truly exist. Chatbot AI as it stands is still just the sum of its parts, or in this case, the sum of its sources. It can take data from the internet, from human works and rearrange them and present them in a user-friendly form - but that isn’t the same as creativity. 

Expect a future where chatbots have become a valued tool to which a lot of the heavy lifting across industries is delegated, but where the human touch is still needed to provide a guiding hand.