人工智能变智障？谷歌版“ChatGPT” Bard首秀大翻车，一夜市值蒸发7000亿元！（附视频&摘要稿）

Google owner Alphabet Inc fell by the most in more than three months after a demonstration of its new artificial intelligence chatbot, Bard, sparked concerns that the tech giant has lost ground in the race for the future of internet search.

Google has been under pressure since developer OpenAI launched its wildly popular chatbot, ChatGPT, which many in the tech industry tout as the next generation of search. On Tuesday, Microsoft Corp., which is investing billions in OpenAI, unveiled a new version of its Bing search engine and Edge browser incorporating technology from the AI startup. On Wednesday, Google hosted a news conference in Paris where it shared more details about its progress integrating artificial intelligence into search.

Investors were largely underwhelmed by the demonstration. In one instance, Bard was asked about new discoveries from the James Webb Space Telescope. In one of its responses, Bard said the telescope was used to take the first pictures of a planet outside the Earth’s solar system — but NASA says those were actually taken by a different telescope.

For now, Bard is available only to a limited number of trusted testers, and OpenAI’s ChatGPT has also been found to deliver inaccurate or outdated responses. Yet investors are keenly attuned to any threat to Google’s search business, which remains its lifeblood.

“That is why you see such a reaction, because this is the money generator, the cash cow in Alphabet’s portfolio,” said Mandeep Singh, an analyst with Bloomberg Intelligence.

Google said in a statement that Bard’s response “highlights the importance of a rigorous testing process.” The company said it will combine external feedback with its own internal testing to ensure Bard’s responses “meet a high bar for quality, safety and groundedness in real-world information.”

Alphabet declined 7.4% to $99.67 at 1:09 p.m. New York time on Wednesday. Earlier, the shares fell as much as 8.9%, the biggest drop since Oct. 26.

Last year, Google declared a “code red” in response to ChatGPT’s release — a move akin to pulling a fire alarm internally, which sent the company’s AI engineers scrambling for a response. Bard’s imperfect performance on the public stage suggests the company may have felt pressure to show off the technology before it was ready, Singh said.

“They did this in haste,” Singh said. “You don’t expect it from a company that is so dominant, and really has always been able to fend off any challenges as far as their core search business.”

While Microsoft won the narrative this week, Google’s longstanding investments in artificial intelligence will ultimately pay off, said Gene Munster, co-founder and managing partner at Deepwater Asset Management.

“Right now the snapshot is: advantage Microsoft,” Munster said. “However, we still think the long-term advantage should go to Alphabet given the resources it has put into AI over the past six years.”

Let us start off with the year so far, are you waking up every morning energized about what's happening in A.I. or are you yourself losing sleep because of the potential concerns? When we consider what's happened with China GDP T over the last few months, we really do need to take.

It's about the hasty miss. The speed before safety.

That's really kind of driven the the release of Chuck Beatty into the into the worlds. We know that researchers have easily gotten around the preventative content filters of and beauty to produce a cyber search, security threats such as new strands of polymorphic malware. They've been able to create a fraudulent phishing campaigns and expose dealers. We've also seen the potential for Chachi, Beatty and other large language models to, in a sense, create fire hoses, disinformation and propaganda that can flood the digital public square with lies and alternative facts.

And so right. I would say there there's a lot of exciting things going on, but there's a lot of reason to for us to be concerned with showing on the screen just how rapid the rollout is from research to development to new product launch across a number of names and platforms in the wall today. I know it's been astonishing tracking the headlines of how quickly new products and new names are born.

It's that pace dangerous or is that just the reality of innovation? I would I would hesitate to say it's the reality of innovation. I would say that we really do need to avoid what's been going on, which is this kind of shocking irresponsibility of bringing things to the market. With with increasing speed, we've seen the move move fast and break things attitude before.

But this attitude is is becoming perhaps more of a concern as the scale of the potential consequences become wider and deeper into society.

And I think that we we really need to we need to press the pause button a bit when it comes to the acceleration of the technologies.

So I guess that the debate moves to who should police their CEO.

When you reflect on the Alan Turing Institute's work around this area, your work in ethics. What's interesting is how many papers have been published by the likes of open A.I.

guidelines that the industry themselves have put forward.

Should they be the ones to do that? Or is there a need for very quickly a conversation around third party oversight? We absolutely need third party oversight when it comes to these systems, especially these highly impactful systems that will affect large populations. Now, that's not to say that the practices of the innovators in designing, developing and deploying these systems shouldn't involve strong ex ante practices of impact assessment, transparency and accountability of bias mitigation.

But these things simply must be subject to the oversight of third party bodies. Now regulators and perhaps certification schemes. David, what is the single biggest ethical consideration around the use of A.I.

in the real world today? I mean, if we're if we're thinking about these large language models and foundation models, we have to recognize that what these systems are doing is merely drawing on.

Billions of of of parameters that that that come from human behavior, that can have discrimination embedded in it, that can be that can be racist and sexist and can can in fact ultimately be re replicated and amplified in systems. And so we really need to be aware that if we do not approach these these systems with a sufficient bias mitigation mechanisms, we will simply replicates deeper patterns than just. Right.

No.