The hype about OpenAI’s AI chatbot ChatGPT apparently caught Google on the wrong foot. Within a few weeks, ChatGPT developed into a crowd puller and broke through the user mark of 100 million faster than any consumer application before. According to media reports, Alphabet CEO Sundar Pichai had issued a “red alert” in the meantime, but Google’s response is now coming – in the form of an announcement. As Pichai writes in the Google blog, the in-house experimental conversational AI will be called “Bard” and will combine “the breadth of the world’s knowledge with the power, intelligence and creativity” of Google’s language models. The most important part of Bard is Google’s well-known chatbot LaMDA, which caused a sensation in mid-2022 because one of Google’s software engineers was convinced that LaMDA had developed consciousness.
Bard is set to become Google’s answer to ChatGPT.
Bard relies on information from the Internet to provide “fresh, quality answers.” Pichai thus suggests that Bard can also provide answers to current developments. ChatGPT’s “knowledge”, on the other hand, ends at the end of 2021 because the AI was only trained with data up to that point. As an example, Sundar Pichai cites Bard explaining to a nine-year-old what the latest discoveries of the James Webb Space Telescope are.
As an example of Bard’s access to current data, Google cites the latest results from the James Webb telescope.
You can’t try Bard yet. Google plans to open up its chatbot to “trusted testers” first, before opening Bard to the public in the coming weeks.
The first version should be based on a “lightweight” model version of LaMDA in order to get by with comparatively little computing power and thus serve more users. Like ChatGPT, Bard is being offered for free during the testing phase to gather as much feedback as possible.
When presenting the latest quarterly figures, the Alphabet boss had already announced “big, AI-driven leaps” in the search engine and other services.
This message is continuously updated.
(vza)
