Senior Vice President at Google , Prabhakar Raghavan , has warned the public about a little - know trouble with hokey tidings ( AI ) chatbots : they " hallucinate " .

Google is set to set in motion its own rival to OpenAI’sChatGPT , a language manikin that can answer your interrogation and queries . Named Bard , the chatbot will roll out out to the populace in the coming weeksaccording to Google CEO Sundar Pichai .

Ahead of the launch , Google demonstrated the superpower of Bard in a promo picture . alas , people notice that the chatbot – a surmount - down translation of their Language Model for Dialogue Applications ( LaMDA ) whichconvinced one engineer it was sentient – came up with incorrect assertion about theJWST .

Bard take that the JWST was the first telescope to take picture of planets outside of the solar arrangement , even though the first look-alike was create by the Very Large Telescope in 2004 , almost two decade before JWST began its work .

As well as being a notecase - ache for investor ( the error wiped over$100 billion off shares ) , the job is a headache for Google and other AI chatbot creators .

Speaking to German newspaperWelt am Sonntag , Raghavan admonish that users may be delivered complete frill by chatbots , despite solution seeming coherent .

" This eccentric of artificial intelligence we ’re verbalize about can sometimes lead to something we call hallucination , " Raghavan told Welt Am Sonntag . " This is then give tongue to in such a way that a machine delivers a convincing but all fictitious reply . "

For example , when ask what is the record for crossing the English Channel on metrical unit , ChatGPT claims that several unlike people have set that record ( taking between four and 12 hours ) .

According toco - beginner of Got It AI , Peter Relan , ChatGPT hallucinates about 15 or 20 percent of the meter .

“ The Francis Scott Key here is to witness out when it is , " hetold Datanami,“and make certain that you have an alternate answer or a response you hand over to the substance abuser , versus its hallucination . ”

Despite teasing that the launch of Bard will happen " soon " last calendar week , no further information has been released . Raghavan told Welt Am Sontag that they want to test the bot on a large scurf so that they can be assure that it is apply out factual selective information .

" Of of course we find the urgency , but we also feel the corking responsibility , " Raghavan read . " We by all odds do n’t want to mislead the world . "