It’s finally there! Bard, Google’s Generative AI answer to ChatGPT, is finally available in new countries and new languages, including French! The Semactic Team took the opportunity to test the new Google Language Model, and compare it to the current superstar, ChatGPT.
What's new ?
This latest Bard update is a big one. Here are the most important higlights and features:
- It’s probably the hottest news: Google Bard, which was previously only available in English and in the US and UK, is now available in more than 40 languages.
- Google is now able to read its answers out loud, which can be really useful if you’re looking for the right pronunciation of a word for instance. This feature is available in all the new languages.
- You can now pin your conversations en rename them to revisit them more easily afterwards.
- Good new for the devs around! It’s now possible to export Python code to Replit
- The answers provided by Google Bard can be shared with your friends or colleagues
There are extra new features that are currently still exclusive to the English version of Bard:
- It’s now possible to adapt Bard’s tone and style to five different options: simple, long, short, professional or casual.
- The use of pictures in prompt through Google Lens. This can help you understand better the picture you see, or help you generate a caption, for instance.
What we liked the most
There are a lot of new features we enjoyed in this update of Google Bard.
- The possibility to read the generated text out loud. It’s always handy!
- The possibility to share Bard’s answers. However, the conversations can’t be shared (yet), and we found it a bit disappointing. It’s always better to get the context behind a simple answer.
- The pin and rename feature
- The different suggestions for every answer generated. It’s immediate and automatical. In ChatGPT, you have to regenerate your request to get another answer. Bard makes it faster!
- The way information is structured with bullet points.
- Bard sometimes provides more complete and exhaustive answers than ChatGPT. It reminded us of Wikipedia. Here’s an example here with “Best running snickers”
What we liked less
- It’s not easily possible to change the input and output languages. It’s possible to directly provide the wished output language in the prompt, but it doenst’ always work very well. In the example here below, we asked Bard to make a short summary in English of the article Google wrote about its latest update, and it first replied… in French! Also, Bard ended up talking about himself using “I”, which wasn’t the point of the summary.
- Talking about language problems, we asked Bard for the best manuals to learn Italian, and it replied to us with a selection of English books, while we’re french native speakers (and we made the request in French).
- Sometimes, Bard would just provide us with wrong or inacurrate information.
- Bard rarely mentions its sources, and does it pretty randomly, or following a pattern we haven’t identified yet.
Bard has really good features we enjoyed, and some we liked less. It’s pretty hard to draw any conclusion right now as Bard is still in development.
In any case, we’re going to keep using and testing it as it (hopefully) keeps evolving, as Bard will be at the heart of the SGE (Search Generative Experience) and will thus be a key element in tomorrow’s SEO.