Google Rolls Out its Bard Chatbot to Battle ChatGPT

Google is not used catching up in either artificial intelligence or search, but today the company is scrambling to show it hasn’t lost its lead. It launches the launch of a chatbot called Bard to do battle with the sensationally popular ChatGPT.

Bard, like ChatGPT, will respond to questions on an almost inexhaustible range of topics and discuss them with a sometimes human-like understanding. Google showed WIRED several examples, including requesting activities for a child interested in bowling and requesting 20 books to read this year.

Bard is also like ChatGPT in that it sometimes makes things up and acts weird. Google published an example where the name of a plant suggested for indoor growing was incorrectly given. “Bard is an early experiment, it’s not perfect, and things will occasionally go wrong,” says Eli Collins, a vice president of research at Google who works on Bard.

Google says it made Bard available to a small number of testers. Starting today, anyone in the US and UK can apply for access.

The bot will be accessible through its own web page and separate from Google’s regular search interface. It offers three answers to each question – a design choice intended to convince users that Bard generates answers on the fly and can sometimes make mistakes.

Google also offers a recommended search query for a traditional web search under each Bard answer. And users will be able to provide feedback on their answers to help Google refine the bot by clicking a thumbs up or thumbs down, with the option to provide more detailed feedback.

Google says early users of Bard found it a useful aid for generating ideas or text. Collins also acknowledges that some have successfully tricked him into misbehaving, though he hasn’t specified how or exactly what restrictions Google tried to impose on the bot.

Bard and ChatGPT show tremendous potential and flexibility, but are also unpredictable and are still in the early stages of development. This poses a conundrum for companies hoping to gain an advantage in advancing and utilizing the technology. For a company like Google with large established products, the challenge is particularly difficult.

Both chatbots use powerful AI models that predict the words that should follow a given sentence based on statistical patterns gleaned from massive amounts of text training data. This turns out to be an incredibly effective way of mimicking human answers to questions, but it means the algorithms sometimes make up facts, or “hallucinate” them — a serious problem when a bot is meant to help users find or search information on the web .

ChatGPT-style bots can also regurgitate prejudice or language found in the darker corners of their training data, such as race, gender, and age. They also tend to reflect the way a user addresses them, making them easy to pretend to have emotions and prone to saying strange and inappropriate things.