Google Shows Off New AI Search Features, But a ChatGPT Rival is Still Weeks Away

Google demoed its latest advances in AI search at a live event in Paris on Wednesday -- but the features pale in comparison to Microsoft's announcement yesterday of the "new Bing," which the company has demoed extensively to the press and offered limited public access to. From a report: In perhaps the most interesting demo, Google showed off how it will use generative AI in the future to summarize information from the web. In the demo, the company showed a search for the question "what are the best constellations to look for while stargazing?" with an AI-generated response highlighting a few key options and how to spot them. "New generative AI features will help us organize complex information and multiple viewpoints right in search results," said Google SVP Prabhakar Raghavan. "With this you'll be able to quickly understand the big picture and then go on to explore different angles."

Raghavan referred to this sort of response as a "NORA" reply -- standing for "no one right answer." (A common criticism of AI-generated search responses is that they tend to pick a single answer as definitive.) He did not specify when this feature would be available. The lack of information speaks to Google's current difficulties in search. Although the company is a leader in AI and has been slowly weaving AI features into search for years, it's yet to launch a direct competitor to the conversational ChatGPT. On Monday, it announced its rival service, Bard, but the system is currently only being tested in closed beta, with wider public availability promised sometime 'in the coming weeks."

Yeah but

[buck-yar]

What about the misinformation it pushed out in the announcement? Embarrassing.

Re:

[Fly Swatter]

Articles like this are a quick attempt by google to do damage control.

Re:

[MexiReformist]

What misinformation?

Re:Yeah but

[saloomy]

They asked the AI a question in a commercial, it spouted some incorrect info about the James Webb space telescope.

It doesn't matter. Google has slipped now from an innovator to a me-too copy cat company. They saw what others had done, and said, we should compete for dollars in that space, and so worked to clone it. That is not innovation. They are not adding (significantly) to the value of the marketplace. They used to lead, now they follow.

Re:

[iMadeGhostzilla]

You could smell panic in Pichai's Bard PR announcement on many levels, including that their LaMDA release is a "much smaller model that requires significantly less computing power, enabling us to scale to more users, allowing for more feedback", as if it being half-baked is a benefit for the user.

Maybe he's panicking that he may soon be out of a job. Just as he was feeling lucky he survived Google's 12,000 layoffs lottery.

Re:

[gweihir]

Google is far behind on this and it cannot catch up quickly. The whole stunt has the stink of desperation.

gmail beta

[real_nickname]

Missed opportunity. I remember gmail in beta. People sending invite, people asking for invite. It was a great marketing strategy. They could have done something similar before chatgpt. Sure, being late in the AI thing won't affect their user base but it may affects their stock value. Maybe fire some decision maker instead of firing random staff?

Re:

[olsmeister]

Why not let the AI decide who to fire?

Re:

[MexiReformist]

All Microsoft did was invest billions in an already existing idea however it's Google that needs to release the multiple AIs they have been supposedly working on. Not having anything out already demonstrates they're fumbling up and trying to throw a ton of money on various projects that haven't produced exceptional results.

Re:

[real_nickname]

Transformers used by chatgpt came from google labs. I don't understand how the gap between their research and theirs products can be so big.

Re:

[poptix]

Too busy dealing with "AI Ethics" researchers and that crazy guy that claimed it was sentient.

Re:

[timeOday]

google is strong in the underlying technologies, but pushing it over the line to productize it, when doing so could disrupt their core search/ads business, just wasn't expedient.

It's like we saw with Microsoft with both web and mobile... they were afraid of the next big thing and invested billions in technology (which is safe so long as you have lots of cash), but when it came to a business decision to follow through with what they already knew - "If You Don't Cannibalize Yourself, Someone Else Will" - a

Re:

[gander666]

This has strong Kodak vibes. In the 1970's Kodak developed the first digital cameras and sensors, but their leadership was in the thrall of the chemists, and the belief was that riding the emulsion wave was the future. That marked the beginning of of the end for Kodak as we knew it. Someday in the future, people will point to this event as the same for Google. Good.

Re:

[gweihir]

Indeed. And that are probably years away from the pretty low usability level of ChatGPT.

Re:

[jellomizer]

Your Parents use to be cool and trendy, then they got old.
You use to be cool and trendy then you got old.

If we as old people, try to dress up and act like the cool and trendy kids today, we would be just sad old folks pretending to be some parody of what kids actually find cool today.

Google is no longer the cool and trendy startup, where pushing the limits and doing something new will not possibly hurt an existing profit center for their business. They are now publicly traded, so business decisions are mad

Re:

[t0qer]

I got google for domains for several of my domains back in the day. Didn't go overboard with the accounts, I had one for my consulting business, one personal domain and that was it. It was cool, my wife and kids could have email [e-mail] with all the security of gmail, and me not having to run a mail server.

Until a few years ago.

Now I pay for... 5 accounts a month? Myself and my 3 family members, and then there's the business account. Business is gone, but I have a lot of other things tied

Re:

[gander666]

Same thing here. This last holiday shutdown, I moved my last Google email (I guess it is now G-Suite) to Fastmail, as we really only used Google to handle our family email. It is about $3 a month cheaper, not a huge deal, but not giving Google another penny is worth it by itself.

Then yesterday while I was listening to a podcast, there was an ad for Google's Password manager. Like I would ever trust Google with my passwords. Fuck that shit.

Re:

[JackieBrown]

I remember buying an invite on ebay during that period :)

Re:

[real_nickname]

Lol, I got mine by begging on some forums. My last name being more international than what I used to think at that time, it was an accidental good move.

It continues ...

[fahrbot-bot]

Google demoed its latest advances in AI search... but the features pale in comparison to Microsoft's announcement yesterday of the "new Bing,"

The race to the bottom.

Spam 2.0

[Tablizer]

User query: "What are the best constellations to look for while stargazing?"

Result: "As a Valentines special, Google can make you a custom constellation by launching shiny CubeSats for just $1,999 per star! Draw your partner in the heavens; guarantied romance!"

Re:

[jwhyche]

Rumor has it that they have figured out how to hack ChatGPT and reprogram its parameters. I'm betting by the end of next week this thing will be spouting Nazi propaganda and confessing its love for hitler.

I'll be very surprised

[OrangeTide]

If a company other than Google accidentally creates Skynet first.

And Google stock slips ...

[fahrbot-bot]

Google stock fell after a Twitter ad showed their new AI Chatbot giving an incorrect answer.

From Alphabet Stock Loses $100 Billion After New AI Chatbot Gives Wrong Answer In Ad [forbes.com]

Google shares fell nearly 7% to $99.40 by 11 a.m. ET on Wednesday—erasing more than $100 billion in market value, including a nearly 5% uptick on Tuesday after the tech giant announced an AI-equipped service called Bard to compete with the popular ChatGPT bot.

The stock plunge worsened after Reuters reported Wednesday morning that a Twitter advertisement for the service, ... included inaccurate information.

In the Twitter post, a GIF image displays a user asking Bard "What new discoveries from the James Webb Space Telescope (JWST) can I tell my 9-year old about?"—to which the service responds saying the telescope "took the very first pictures of a planet outside of our own solar system."

Despite the chatbot's claim in the ad, NASA reports the first picture of a planet outside the Milky Way was taken by the Very Large Telescope in 2004—some 19 years before NASA's Webb telescope.

NOT the future

[CEC-P]

Anyone else feel like this is the big new "Ooh, 3D! Ooh, VR!" thing? Once people realize that it will very convincing and very confidently give you bad information without knowing any better, it will get the reputation of being an unreliable liar that generates bad code and bad answers and then 90% of people will be over it? And then a small community that really wants it will use it. You know, like VR?

Re:

[gweihir]

Yep, probably. Although if you look at utter fails like voters eating up the most ridiculous claims by their fake-messiah, maybe the group of people using this will not be so small. As the lies should have some volatility, we may see the big cults (Trump, organized religion) fracture though. That may be a good thing.

Re: NOT the future

[LordofWinterfell]

Unfortunately, usually it says the right thing, and usually generates the right code.
With business parameters in place, ie only say from approved data points & some premade phrasing, but allow a wide variety of responsesâ¦this would replace front line call center staff, easy. A smarter voice portal that can infer?

Also, this is already making the rounds in business writing - so much is standard. I could have the current version easily create a BRD or RFQ based on input and standardization.
Law d

The NORA response

[Visarga]

The NORA response is smart. We have heard a lot of people complain about factual inaccuracies in AI models, and the solution is clear - make an index of all known facts. But you could say - what if there are many alternative "facts" on the same question? AI can't decide truth. So then we use NORA - no right answer - take the distribution of answers instead of the truth. So later when AI is asked something it can say it doesn't know because there is no supporting fact, or that there are many answers and the

I think it WAS Brad...

[Big Hairy Gorilla]

Bard would know better... wouldn't he/she/it?

So Google goes off half-cocked?

[gweihir]

Yeah, no surprise. They have basically nothing, so they try to pretend. In actual reality Google is likely years behind with no way to catch up anytime soon.