Google Bard is Switching To a More 'Capable' Language Model, CEO Confirms

People haven't exactly been impressed in the short time since Google released its "experimental conversational AI service" Bard. Coming up against OpenAI's ChatGPT and Microsoft's Bing Chat (also powered by OpenAI's GPT-4) users have found its responses to not be as knowledgeable or detailed as its rivals. That could be set to change, however, after Google CEO Sundar Pichai confirmed on The New York Times podcast "Hard Fork" that Bard will soon be moving from its current LaMDA-based model to larger-scale PaLM datasets in the coming days. From a report: When asked how he felt about responses to Bard's release, Pichai commented: "We clearly have more capable models. Pretty soon, maybe as this goes live, we will be upgrading Bard to some of our more capable PaLM models, so which will bring more capabilities, be it in reasoning, coding." To frame the difference, Google said it had trained LaMDA with 137 billion parameters when it shared details about the language-based models last year. PaLM, on the other hand, was said to have been trained with around 540 billion parameters. Both models may have evolved and grown since early 2022, but the contrast likely shows why Google is now slowly transitioning Bard over to PaLM, with its larger dataset and more diverse answers.

We clearly have more capable models

[coop247]

Just didn't want to show off.

I clearly have a very attractive girlfriend. No you can't meet her.

Re:

[cstacy]

I clearly have a very attractive girlfriend. No you can't meet her.

Also, I have up to 540 billion parameters in my model.

Idea

[backslashdot]

Train it on the slashdot comments and article history. What would happen?

Re:

[Dictator For Life]

CowboyNeal becomes responsible for much of modern civilization, and MEEPT!! is unmasked.

Re:

[real_nickname]

Chatgpt hasn't been unfortunately or is lying: "backslashdot" is not a known username or registered account on Slashdot. It's possible that the username was used briefly or may have been created on a different platform altogether.

Re:

[MightyMartian]

In Soviet Russia, AI-generated eye-less humanoid monstrosities own you!

Re:

[PPH]

article history

The first dupe it finds, it gets stuck in an infinite loop.

Re:

[Tough Love]

Hot grits becomes a fad diet.

I want to go back to 2020

[Rosco P. Coltrane]

At least during the pandemic, we didn't get to hear about fucking OpenAI, fucking ChatGPT, fucking Bing or fucking Bard every 10 fucking minutes.

Re:

[MobileTatsu-NJG]

"I wanna frequent a tech news site but I somehow don't want hype to come along with that!"

Ironically ChatGPT could probably do a decent job of filtering the stories you don't want to see for you.

Re:

[HiThere]

Not yet.

Re: I want to go back to 2020

[DivineKnight]

Agreed. Marketing for these companies is sure pushing hard.

Re:

[Dutch Gun]

Yes, but you'd instead be reading about all the crypto hype. I'm not sure I could stomach more of that.

Re:

[linuxguy]

OpenAI/ChatGPT and AI in general is now a much bigger deal than anything we have seen in technology for decades. It will have the tremendous impact on our lives. Positive or negative, that is yet to be determined.

Re:

[Rosco P. Coltrane]

Positive or negative, that is yet to be determined.

Like everything, it'll be positive in the long run. But here's the thing: this one will wreak so much havoc so quickly that society will be destroyed before anything positive has a chance to come about.

Case in point: one of our engineers at work played with GPT-4 and asked it to generate what was essentially a week of work for a junior programmer in 30 minutes. Sure enough, there are now talks of firing a few employees and putting a few hires on hold higher up.

When I reported this to a business owner friend

Re:

[antdude]

So, U'd rather hear about COVID19 every 10 mins.

Re: The Search Engine Sucked, WECGR with AI?

[Blackeneth]

ChatGPT summarizes your comment thusly:

"The text argues that the quality of search results from search engines like Google has consistently degraded since 2010 due to factors such as mobile devices, SEO spam factories, and the influx of cheap cellphone users. This has resulted in a shift from quality data to ad-supported spam, which is now being catered to non-English speaking countries with impaired AI language models. The author suggests that new digital infrastructure is needed to establish defined borde

Fire Sundar Pichai

[S_Stout]

He has failed at everything in his time there. He was handed a company at the top and now they are playing catch up. He has zero accomplishments. Fire him. Now. Do it right now before he fucks up again, which will be next week and every week after.

Re:

[ap7]

It is less Pichai and more the woke culture that is sweeping across many tech orgs.

More energy was being spent on discussing and fighting over issues completely peripheral to the organization's core objectives.

There was a great piece on Google once. The decline actually began well before Pichai. Like Yahoo and many others, they have just stopped making great-to-use products that people needed.

https://www.vox.com/2015/10/23... [vox.com]

I wanted to try ChatGPT...

[rlwinm]

So I wanted to try ChatGPT and it wanted a phone number. A cellular phone number. I refused to give it one and used a friend's account instead. I was underwhelmed to say the least with what ChatGPT could do. If I can try this nonsense without lots of signup pain I may but I am thinking I will be underwhelmed with the Google one too.

I tried Google Bard. It is very dumb.

[linuxguy]

I frequently use ChatGPT 4 during the day as a software engineer. It is very useful. Version 3.5 was pretty good. Version 4 is much better. I tried Bard. Gave it a couple of simple programming tasks. It just plain refused to do them. Said it wasn't designed for that sort of thing. Google is far behind at this point. I do want them to get better. We don't want a monopoly in this field.