Google’s search chief has warned against always relying on artificial intelligence chatbots to produce accurate results. On Saturday, Prabhakar Raghavan told Welt Am Sonntag that they could sometimes give false but convincing answers.
“The type of artificial intelligence that we are currently discussing can occasionally lead to what we call hallucination,” he told a German newspaper. “It was then expressed so that a machine provides a convincing but fictitious response.”
Google used a presentation on Monday to unveil Bard, an AI chatbot that it hopes will compete with ChatGPT. However, an advertisement for Bard depicted it as providing an incorrect answer to a question about the James Webb Space Telescope.
Alphabet, Google’s parent company, saw its stock plummet after the error and rising concerns about ChatGPT’s threat to Google’s search dominance. Raghavan told Welt Am Sonntag that Google was concerned that “They have a strong desire to make Bard available to the public, but we also have a strong sense of responsibility. We certainly do not want to mislead the general public.”
“We’re thinking about incorporating these options into our search functions, especially for multiple-choice questions,” he added. Raghavan’s remarks come after some chastised Google and Alphabet CEO Sundar Pichai for Bard’s “hurried and bungled” launch.
Former Alphabet AI division Google Brain research engineer Maarten Bosma tweeted that the presentation demonstrated the company’s lack of commitment to AI. Google did not immediately respond to an Outsider’s request for comment after regular business hours.