Google's CEO said, 'We will update Conversational AI Bard in the near future,' and the possibility that Google Assistant and Bard will be integrated



In February 2023, Google announced a conversational AI service `` Bard '' that uses conversation-specific AI `` LaMDA '', but Google declared `` code red '' with the appearance of ChatGPT, which became explosively popular. It has been reported that Google employees are dissatisfied with Bard's hasty announcements , and it has been reported that Bard is less capable than ChatGPT , and it is often pointed out that Bard is immature. It has been. Google CEO Sundar Pichai responded to an interview on the New York Times podcast on March 31, 2023, saying, ``In the near future, we will have a better model for Bard,'' more efficient and multifunctional. unveiled a new model.

Google CEO Sundar Pichai on Bard, AI 'Whiplash' and Competing With ChatGPT - The New York Times

Google CEO Sundar Pichai promises Bard AI chatbot upgrades soon: 'We clearly have more capable models' - The Verge

Google Bard will soon switch to a more powerful language model, CEO confirms | Engadget

Google reorganization in Assistant follows Bard launch, memo says

Google announced the conversational AI service `` Bard '' on February 8, 2023, but the animated image introducing Bard gave an inaccurate answer, and the stock price of Alphabet, Google's parent company. A total loss of about 120 billion dollars (about 15 trillion yen) was given, and Bard read the text generated by AI and misunderstood that he had already finished the service, Bard ChatGPT Problems have often been pointed out, such as beingrumored to be trained with data from . Also, according to CNBC's interview, in the internal forum used by Google employees, there were many critical voices about Bard's announcement, such as 'too hasty', 'failure', and 'not like Google', and Google and Mr. Pichai's choice There were also comments directly criticizing the

Google's chat AI ``Bard'' gave an inaccurate answer, causing Google's market value to drop by more than 15 trillion yen - GIGAZINE

Pichai did not directly respond to his dissatisfaction with Bard, but appeared on the New York Times podcast on March 31, 2023, and revealed a major update to Bard in an interview.

According to Mr. Pichai, he plans to upgrade the language model 'LaMDA', which is the basis of Bard, to a higher performance 'PaLM' model. ``The Bard runs on a lighter and more efficient version of the LaMDA, but it's like putting a powered up compact Civic into a race against a more powerful car,'' Pichai said. , PaLM seems to be more versatile than that, and has a higher ability to handle tasks such as coding and math questions, which Bard was not good at, in addition to common sense reasoning. 'PaLM allows us to better understand and respond to prompts for multi-step word and math problems,' said Jack Crouchik, senior product manager at Google. We are always aiming for a balance that improves efficiency while adding

PaLM is a pre-trained language model with about 540 billion parameters, which is quite large compared to LaMDA's training parameters of about 137 billion and the language model 'GPT-3' used for ChatGPT of about 175 billion. parameters.

Mr. Pichai was cautious about Bard's low ability, saying, ``It was important not to produce a more functional model before fully confirming that AI could handle the task well.'' I'm saying that it's because I was taking . He also acknowledges that his reliance on LaMDA has made Bard's capabilities smaller, and he slowly transitioned to PaLM after receiving more user feedback and developing his own analysis of Bard's safety and quality. I promised to do it.

In addition to updating Bard, Google has also revealed that it is working on a project that fuses Bard and Google Assistant. According to CNBC, Google's vice president and head of Google Assistant, Sissi Xiao, has sent a memo to employees titled 'Google Assistant and Bard Team Changes,' suggesting the Google Assistant team will focus on Bard. He said he did.

In response to the New York Times report that Google declared 'Code Red' following the release of ChatGPT, Pichai said that he and Google co-founders Larry Page and Sergey Brin were among the code He denied that he had ever issued Red internally, saying, ``I think there was someone in the development team who was in a scrambled state who made such an expression.''

With the release of ChatGPT, Google declares ``code red'' and reassigns teams to respond to the threat AI chatbots pose to the search business-GIGAZINE

In addition, Mr. Pichai said that more than 1,000 people had signed a letter calling for development to be stopped for fear of losing control of AI , and that the AI technology ethics group had told the Federal Trade Commission that the deployment of OpenAI's generative AI was 'public. ' Regarding 'concerns about AI,' such as requesting an investigation because it poses a risk to safety, he said, 'I think it's important to listen to concerns in this field. No one knows all the answers.' No company can get it right, but AI is too important an area to regulate.”

in Software,   Web Service, Posted by log1e_dh