In a newspaper interview published on Saturday, Google’s search engine boss warned of the pitfalls of artificial intelligence in chatbots, while Google’s parent company Alphabet is fighting for competition with the hit app ChatGPT. This type of artificial intelligence we’re talking about can sometimes lead to so-called hallucinations,” Prabhakar Raghavan, Google’s senior vice president and head of Google search, told German newspaper Welt am Sonntag.
ChatGPT’s ability to accurately and truthfully answer a question could save users the time they normally spend scrolling through links on Google to find the same information. In addition, Microsoft has integrated an improved version of ChatGPT into some of its products such as Teams or Bing to have an added value over the competition.
Not to be outdone, Google has ramped up work on Apprentice Bard, one of its ChatGPT answers.
Anonymous employees told the media that Apprentice Bard’s implementation is similar to ChatGPT (the bot provides detailed answers to user input and questions). However, instead of using the GPT 3.5 code developed by Microsoft, Bard uses LaMDA (or Language Model for Dialogue Applications), Google’s language model for dialog applications.
LaMDA is Google’s system for building chatbots based on its most advanced large language models, so called because they mimic speech by taking billions of words from the web. The major search brand publicly showcased LaMDA at the 2021 edition of Google I/O, their conference for developers aiming to improve their artificial intelligence conversation assistants and make conversations more natural. The company already uses similar language model technology for Gmail’s Smart Compose feature or for search engine queries.
Last week, Google announced its AI Bard chatbot. But the bot didn’t get off to a good start, as experts noted that Bard made a factual error in his very first demo.
A GIF shared by Google shows Bard answering the question: What new discoveries from the James Webb Space Telescope can I tell my 9-year-old about? Bard offers a three-point enumeration, including one that says the telescope made the first-ever images of a planet outside our own solar system.
Bard’s very first reply contained a factual error
However, a number of astronomers on Twitter pointed out that this was wrong and that the first image of an exoplant was taken in 2004 – as the NASA website states: This is not to make my asshole (actually, if at all) and I’m sure Bard will be great, but for the record, JWST didn’t take “the first-ever image of a planet outside our solar system,” astrophysicist Grant Tremblay tweeted.
Also, employees expressed their anger at the company’s management, especially the CEO, for rushing and not thinking before informing about the release of ChatGPT’s competing chatbot. They felt this was not being managed well and could prove to be a costly mistake in the long run.
On Memegen, a popular internal forum, they called this communication strategy rushed, sloppy and not Googley when memes about the situation started appearing online.
While Google employees often turn to Memegen for humorous pokes fun at the company’s quirks and missteps, posts after Bard’s announcement took on a more serious tone, even going straight to Pichai.
Dear Sundar, Bard’s launch and shots were rushed, botched and short-sighted,” read a meme that included a photo of Pichai. Please return to a long-term perspective. The embassy received many upvotes from employees.
Sundar and his leadership deserve an NI Perf,” said another highly rated meme, referring to the lowest category in the company’s employee performance appraisal system. They’re oddly short-sighted and un-Googleish in their quest for sharpness.
Some people think it’s a poor attempt by the CEO to tell the world that he’s still in the running for his own offering. Don’t forget that Microsoft has already made its Bing chatbot public and now all eyes are on Google’s release.
Google boss urges caution
In this context, the head of the Google search engine Raghavan advises caution when using chatbots with artificial intelligence: This type of artificial intelligence we are talking about can sometimes lead to so-called hallucinations, said Google search engine boss Prabhakar Raghavan WELT AM SONNTAG.
He added: This is then expressed in terms of a machine providing a convincing but entirely fictitious answer. According to him, reducing this to a minimum is one of the basic tasks.
Microsoft recently announced plans to boost its Bing search engine with an improved version of ChatGPT to better compete with Google. Bard, the counterpart to ChatGPT developed in Google’s internal kitchens, is currently being tested by some users. We’re also thinking about how we can integrate these options into our search capabilities, especially for questions that don’t have a single answer,” explains Raghavan.
The group has not yet revealed when Google will publish its chatbot: “Of course we feel the urgency, but we also feel the great responsibility,” said Raghavan. We certainly don’t want to mislead the public.
Google search engine boss Prabhakar Raghavang at a conference in San Francisco in 2018
Some people think Bard’s rushed presentation was a poor attempt by the CEO to tell the world that he’s still in contention for his own offering. Don’t forget that Microsoft has already made its Bing chatbot public and now all eyes are on Google’s release.
On the other hand, Google recently mentioned how exactly the chatbot technology works during its livestream event on Thursday. But during that time, we’ve also seen Microsoft give its presentation and launch its search engine, while sending out invitations to others to come and see how it all works in person.
“The huge language models behind this technology prevent humans from monitoring every conceivable behavior of the system,” Raghavan told WELT AM SONNTAG. But we want to test it on a scale large enough that we’re ultimately happy with the metrics we’re using to verify the answers are correct.
People should always be vigilant when confronted with the results presented to them by an AI-powered chatbot, the Google executive said. It is important that users have the necessary tools to verify the results. This includes disclosure of sources.
We hold ourselves to a very high standard,” Raghavan said. And it is also our goal to be a leader in chatbots in terms of the integrity of the information but also the responsibilities we take. This is the only way to maintain public confidence.
Diploma
Alphabet shares fell more than 9% last week on concerns about ChatGPT’s threat to Google’s core search business and the company’s hasty response to Microsoft’s event. Panicking Bard into the market confirmed the market’s fear of us,” he wrote in a much-loved meme.
For Maarten Bosma, a former research engineer at Google Brain, Alphabet’s AI division, Google’s presentation showed that the company wasn’t taking AI seriously enough: I think Google’s presentation signaled (wrongly or wrongly) that they didn’t really take it seriously.* : It was less than 5 minutes, among other things it was present, it was Paris instead of headquarters, the CEO wasn’t there. Bing, on the other hand, has changed the homepage.
I think the Google presentation signaled (rightly or wrongly) that they don’t really take it seriously: it was less than 5 minutes, it was presented among other things, it was in Paris instead of HQ, the CEO wasn’t there. Bing, on the other hand, has changed the homepage.
— Maarten Bosma (@MaartenBosma) February 9, 2023
Google’s position in the field of artificial intelligence is a well-known concern. At a town hall meeting in December, employees questioned the company’s competitive advantage in AI when ChatGPT became public. Executives responded that the company’s reputation could suffer if it transitioned too quickly to AI chat technology, which is far from perfect.
A widely read meme last week featured an image of a dumpster fire with the Google “G” logo on it. The lyrics read: How everything has felt since last year.
Another meme related to the layoffs announced last month, which resulted in the loss of 12,000 jobs, or 6% of the company’s workforce. Featuring a photo of smiling actor Nicolas Cage, the meme read: 12,000 layoffs increase market cap by 3%, hasty AI pitch depresses it by 8%.
Source: World on Sunday
And you ?
What do you think of the words of the Google boss?
Adds the anger of the company’s employees, who have criticized CEO Sundar Pichai for Bard’s hasty and sloppy announcement, do you think they are motivated by the bug seen by the public, or are they not a cold observation of reality regarding these tools in general ? To what extent?