Google Bard has landed in Europe, including language support for plenty of local languages and even text-to-speech in a select number of them. What can you do with the chatbot and what is it suitable for?
Google released Bard a few months ago, which was made available in experimental form in the United States. Unlike OpenAI’s AI sensation ChatGPT, initial reactions were not immediately thunderous, but the dev team gradually refined the chatbot to be a reasonable competitor to the Microsoft-backed product.
Disclaimer ahead
Google Bard became available today in the Netherlands and the rest of Europe, including support for our language. Those who want to use the bot must be logged in with a Google Account. Then a disclaimer is shown that emphasizes the experimental nature of the tool and how it can make errors.
You can tell from the prompt suggestions in that initial message that Google is positioning the tool as a helpful source of inspiration rather than a valuable professional tool. However, we immediately asked if it could be used for professional purposes, to which it responded with a resounding “yes.” After several positives, Bard concludes with the disclaimer, “I am not perfect and I sometimes make mistakes. However, I am always learning and I will do my best to give you the best experience.”
A somewhat striking difference from ChatGPT, which responds to the same question with a lot more caution: “While I can offer useful information and suggestions as an AI assistant, it is important to understand that, as a language model, I have no professional expertise in specific areas.” OpenAI’s chatbot recommends consulting an expert.
Falls into repetition often, American origin
A few more questions reveal that the Google chatbot very often repeats itself. When talking about the AI assistant’s functionalities, it will always include a similarly-worded warning.
Furthermore, it is notable that the bot still betrays its own American origins. For example, it uses Fahrenheit and miles per hour, and a distinct tone can sometimes be recognized even when set to Dutch, which is often created by a machine translation.
Initially, Bard states that it knows nothing about you, but it asks counter questions to get to know you better. This is in stark contrast to ChatGPT which quickly hits the emergency brake and seems a little spooked to even discuss the issue, spitting out warning after warning about how it really isn’t allowed to know your personal details.
Intended for text, programming code, scheduling
Ultimately, a generative AI chatbot is bound by many limits. An LLM depends on a dataset and as a tool, it must comply with international laws, which limit the information gathering and therefore the overall impact of a chatbot.
Bard is essentially suitable as an option to summarize texts, draft e-mail messages or give scheduling suggestions. Another interesting option is to generate programming code. Beyond that, you’re at the whim of your own creativity and fact-checking.
Also read: European websites do not appear to train Google’s AI models