Google CEO Sundar Pichai acknowledged that there are several features of Bard, its AI-powered chatbot, that they still don’t understand. He spoke, for example, of various emerging abilities that his model has experienced in recent tests, such as reasoning, planning or creativity.
“You don’t fully understand how it works. And yet, you’ve released it to society?” asked Scott Pelley, an interviewer for the CBS, to Pichai. “Yes. Let me put it this way… I don’t think we fully understand how a human mind works either,” replied the Google leader.
Pichai explained that there is an element in his developments that they call “the black box”, which refers to those aspects that they cannot understand. “Sometimes you can’t tell why he said this or why he was wrong. We have some ideas and our ability to understand him improves over time. But that’s where the state of the art is,” he added.
In the interview they mentioned how a Google AI system learned, by itself, the Bangladeshi language. Also of his ability to reflect on the pain that humans feel or issues such as redemption.
Are they just algorithms putting together speeches or is it a more developed intelligence? “I think we need to approach this with humility… I think it’s good that some of these technologies come out so that society can process what’s happening. And start this conversation and debate,” Pichai replied.
Google advocates regulating AI
The CEO of Google said that he is losing sleep over the possibility that this AI could be “very damaging if implemented incorrectly.” He believes that society must hurry to create clear rules for the implementation of AI in the economy. also asked laws punishing abuse and treaties between governments to ensure that these advances are safe for everyone.
Pichai admitted that it is a mistake for companies to get carried away by competitive dynamics. “It is not convenient to launch a technology like this when it is very, very powerful, because it does not give time for society to adapt”said Pichai in an interview with the CBS. He assured that the company is looking for better ways to approach these systems and that, for this very reason, it has delayed new tests.
The Mountain View tech firm is, however, one of the leading players in this race. At times, it is more prudent than OpenAI or Microsoft. Bard, his conversational model, was released in a much more limited test version than ChatGPT. But even Google workers themselves have warned, for example, that the AI chatbot was introduced without considering its ability to generate false or dangerous content.
AI will be more transformative than electricity
“I have always seen AI as the most profound technology humanity is working on. Deeper than fire or electricity or anything we’ve done in the past”added the CEO of Google in the interview with the CBS. So he doubts society is ready for some more powerful AI systems.
There is a mismatch in the rate at which we can think and the speed with which this technology is evolving, he explained. Google is an example of this. As Pichai reflected on it in an interview, he reported that the company is working on updating its search engine. Google will introduce a completely new one powered by AI, just like Microsoft did with Bing.
The new search engine is part of the “Magi” project, detailed the New York Times. The new tool is intended to be more customizable than the current engine. I would have a image generator and an integrated chatbot in the browser itself.