(ORDO NEWS) — AI-based chatbots can pose a threat to users due to their imperfection – they have a kind of “hallucination”, said Prabhakar Raghavan, senior vice president of Google and head of Google Search, in an interview with the German newspaper Welt am Sonntag.
“The kind of AI that we are talking about now sometimes comes to what we call hallucinations.
This is expressed in the fact that the machine gives a convincing, but completely invented answer, ”said Mr. Raghavan.
And he added that one of the fundamental tasks is to reduce the number of such incidents to a minimum.
Google first began to lose ground in web search technologies after Microsoft secured the support of OpenAI for $10 billion, which recently introduced the revolutionary ChatGPT project.
The language AI model responds to requests in a surprisingly human manner: it got to the point that the chatbot was invited for an interview, and he also wrote a thesis.
The search giant recently introduced its own chatbot Bard, which could not be saved from “hallucination” even in a commercial – it gave a clearly unreliable answer to a question on astronomy.
This caused discontent among the scientific community, and the shares of Alphabet, which owns Google, collapsed by 8%.
“Of course, we are aware of the urgency [of solving the problem], but we also feel a great responsibility. And we definitely do not want to mislead the public, ”added the head of Google Search.
—
Online:
Contact us: [email protected]
Our Standards, Terms of Use: Standard Terms And Conditions.