Mountain View, California (dpa) – Google has announced that its AI chatbot Bard, the tech giant’s rival to the Microsoft-affiliated ChatGPT, is being rolled out in 180 countries around the world, but that users in European Union countries will have to wait.
Bard, which now speaks English, Japanese and Korean, is being kept out of the EU as the US company seeks to clarify whether Bard is compatible with the bloc’s legal framework.
Bard is also getting an improved design and results, notably with clear sourcing for its information, a dark mode and the option to export programming code that it has generated, the company announced at its annual I/O developer conference on Wednesday. Language support for 40 other languages is also set to follow soon.
Showcased above all in its ability to answer complex questions, Bard is able to formulate complete sentences in response to queries while providing the usual internet links and sources for its information.
When searching for a bicycle, in addition to advice to pay attention to the suspension, for example, suitable offers from shops are also displayed.
”What’s better for a family with kids under 3 and a dog,” was one of the sample questions put to Bard with reference to two US parks.
To this, Bard gives key information answering the question, links to further information, as well as follow- up questions about things like how much time to spend at certain parks.
Bard is Google’s answer to ChatGPT from the Californian start-up OpenAI, which in turn is closely linked to the software corporation Microsoft through billions in investments.
Google also introduced a new language model called Palm 2 for novel AI functions to compete against OpenAI’s GPT-4. Palm 2 can master more than 100 languages and brings writing, programming and analysis capabilities. Google’s chat bot Bard will also work on the basis of Palm 2.
ChatGPT has drawn much attention since its public-facing version launched last year, sparking debate as to how AI could change various industries.
Bard, like ChatGPT, is based on a large research language model (LLM). OpenAI works with GPT, currently in version 4. At Google, Bard runs on a trimmed-down and optimised version of the LaMDA language model and is to be updated over time with newer, more powerful models.
Google’s arch-rival Microsoft entered into a billion-dollar pact with OpenAI and is bringing AI into its applications across the board. Google has so far held back on this, citing responsible use of the technology.
At Google I/O, the company stood by it. “The only way to be truly bold in the long-term is to be responsible from the start,” James Manyika, who is responsible for social responsibility in the use of
AI at Google, said.
The company sees the danger that the software could strengthen prejudices or be used to produce false information. To protect against this, files generated with the help of Google’s AI are to be provided with metadata so that they can be recognised immediately.