Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Bard’s latest update: more features, languages and countries

blog.google

Post reply on HN
Print view View original

111–120 of 205 posts

Re: Bard’s latest update: more features, languages and countries

#111

Post by kernal »

isoprophlex wrote: post #61
frankohn wrote: post #56

Just tried out Google Bard, now available in Europe, and I have to say, my first experience was rather disconcerting. I threw at it a relatively obscure question about an event in the third book of A Song of Ice and Fire series, involving characters Arya Stark and Sandor Clegane. To my surprise, Google Bard's response, while confident, was a bizarre mix of actual characters and situations from the book series, but th…

Don't bother trying their other models available in GCP either. Their embedding models and their un-RLHFd generic GPT analogues are miles behind the competition. It's incredible how bad google have dropped the ball on this. We had some google people come in at $CORPO_DAYJOB the other day to sell their cloud offerings. The engineer had the gall to say "people ask me, why did google miss the boat on LLMs? I say people,…

LLMs are in their infancy. ChatGPT is bleeding users by an alarming amount. It's incredible how bad OpenAI have dropped the ball on user retention, product integration and revenue.

Re: Bard’s latest update: more features, languages and countries

#112

Post by kypro »

q7xvh97o2pDhNrh wrote: post #82
kypro wrote: post #52

Earlier quoted context omitted.

I was using it for some simple programming tasks the other day. I figured it should be okay to speak to it with my working class mannerisms since it wouldn't have the biases of a human middle-class programmer. Initially it gave me answers that didn't work. Then I realised the API I was asking it to generate code for was updated in the last year so it was basically useless. Then I asked it some more general coding que…

> it went on an unprompted rant about how I should speak to it "politely". Can you share the prompt that led to this? I agree this sounds pretty disconcerting.

Alright... Please don't judge me for how stupid this conversation is.

Also, looking back at it now I probably should have just ignored its attempt to moralise instead of questioning it.

https://ctxt.io/2/AABQPBXNEw

Re: Bard’s latest update: more features, languages and countries

#113

Post by Workaccount2 »

yjftsjthsd-h wrote: post #77
ipaddr wrote: post #49

Earlier quoted context omitted.

Everything google offers is in beta and remains that way for a very longtime. Expect bugs but not regression of core product offerings.

> Expect bugs but not regression of core product offerings. Google's web search has gotten worse over the years. Unless of course we want to say that their only truly core product is ads, in which case I'm skeptical but willing to concede that you may be right.

Assistant has somehow gotten worse too. I'd take 2018 assistant over the current one.

Re: Bard’s latest update: more features, languages and countries

#114

Post by lgas »

wokwokwok wrote: post #98
lhl wrote: post #90

Earlier quoted context omitted.

It's one thing to say to "use code with caution" but it's another thing to pretend to run a calculation and then hallucinate the answer (or hallucinate that it's running code). I just tried out this exact example. ME: "Do you have access to a code interpreter like Jupyter Lab, Colab, or Replit?" BARD: "Yes..." ME: "OK, great, can you execute the code to give me the prime factors of 15683615?" BARD: Prints code block.…

> I've never seen it make up results inside the results of a code execution block like Google Bard does. What? Can chat-gpt run code in a sandbox? I’ve never heard of this before.

They rolled out code interpreter to everyone this week. https://help.openai.com/en/articles/6825453-chatgpt-release-...

Re: Bard’s latest update: more features, languages and countries

#115

Post by minsc_and_boo »

yjftsjthsd-h wrote: post #77
ipaddr wrote: post #49

Earlier quoted context omitted.

Everything google offers is in beta and remains that way for a very longtime. Expect bugs but not regression of core product offerings.

> Expect bugs but not regression of core product offerings. Google's web search has gotten worse over the years. Unless of course we want to say that their only truly core product is ads, in which case I'm skeptical but willing to concede that you may be right.

I'd argue the internet has gotten worse over the years too.

Search results are only as good as the links to content that others make.

Re: Bard’s latest update: more features, languages and countries

#116

Post by munificent »

yjftsjthsd-h wrote: post #77
ipaddr wrote: post #49

Earlier quoted context omitted.

Everything google offers is in beta and remains that way for a very longtime. Expect bugs but not regression of core product offerings.

> Expect bugs but not regression of core product offerings. Google's web search has gotten worse over the years. Unless of course we want to say that their only truly core product is ads, in which case I'm skeptical but willing to concede that you may be right.

> Google's web search has gotten worse over the years.

People keep pointing that out, but few want to acknowledge that the web has gotten worse over the years.

When was the last time you stumbled onto a web page written by a caring human author to accurately explain some part of the world just because they're into it? Those kinds of pages used to be all over the web. If you did a search for some random topic, you'd find, essentially, a non-commercial fansite for that exact topic.

But these days, much of the web is commercialized, fewer users know how to write HTML, and people who want to share content typically do so behind walled gardens like Facebook.

Blaming Google's web search for getting worse is like blaming your tour guide when you have a shitty hike because the forest burned down.

Re: Bard’s latest update: more features, languages and countries

#117

Post by lhl »

ajross wrote: post #95
lhl wrote: post #90

Earlier quoted context omitted.

It's one thing to say to "use code with caution" but it's another thing to pretend to run a calculation and then hallucinate the answer (or hallucinate that it's running code). I just tried out this exact example. ME: "Do you have access to a code interpreter like Jupyter Lab, Colab, or Replit?" BARD: "Yes..." ME: "OK, great, can you execute the code to give me the prime factors of 15683615?" BARD: Prints code block.…

> It's one thing to say to "use code with caution" but it's another thing to pretend to run a calculation and then hallucinate the answer (or hallucinate that it's running code). Not to an LLM, it isn't. You're asking for "reasoning" features, the idea of having a model of what's needs to happen and whether or not the output matches the constriants of the model. And that's not what LLMs do, at all. That Bard attempts…

I think you're making some improper assumptions about my familiarity with the technology. The whole point of adding grounding via external plugins, code interpreters, or retrieval augmentation is to reduce hallucinations - if a block of code and results are going to be presented/represent itself as the execution output, then it better damn well be directly retrieved and not be rendered by the LLM itself if Google can't train their LLM to not mangle the output.

This part is not rocket science and I think anyone that doesn't understand why making output trustworthy has to be the number one priority, will continue to not succeed in the market.

Re: Bard’s latest update: more features, languages and countries

#118

Post by munificent »

buildbot wrote: post #93
ChatGTP wrote: post #70

Earlier quoted context omitted.

Question, do you think this has to do with copyrighted information being in the models ? I know you might not care but I wonder if Google is operating of accounts from the internet while OpenAI has actually ingested the books ?

No. A blog post on the internet is also copyrighted! Google has probably the largest collection of digitized books as well. They have no reason not to. ML networks have been trained on copyrighted data since before 2012.

> They have no reason not to.

Ethics?

Re: Bard’s latest update: more features, languages and countries

#119

Post by yeldarb »

The image functionality is pretty underwhelming.

Some things I tried that it failed at:

* Simple "spot the difference" a young child could do

* Counting chess pieces on a board or coins in a mario screenshot

* Describing a photo of a car on a road

* Evaluating a graph of y=x and one of y=x^2

* Explaining a meme image (said it can't do anything with images of people)

* Taking a screenshot & giving HTML for it (gave random HTML unrelated to image)

* Taking an image and converting to SVG (gave random SVG unrelated to image)

* Describing a photo of a car (it was parked on a road, Bard said it was driving in front of a brick wall)

It was quite good at OCR though even for images that are pretty tough for other models (eg serial numbers on industrial parts).

Re: Bard’s latest update: more features, languages and countries

#120

Post by isoprophlex »

kernal wrote: post #111
isoprophlex wrote: post #61

Earlier quoted context omitted.

Don't bother trying their other models available in GCP either. Their embedding models and their un-RLHFd generic GPT analogues are miles behind the competition. It's incredible how bad google have dropped the ball on this. We had some google people come in at $CORPO_DAYJOB the other day to sell their cloud offerings. The engineer had the gall to say "people ask me, why did google miss the boat on LLMs? I say people,…

LLMs are in their infancy. ChatGPT is bleeding users by an alarming amount. It's incredible how bad OpenAI have dropped the ball on user retention, product integration and revenue.

I'm not discussing, let alone contesting user retention, revenue or integration. I'm making an observation on my perception of raw model output.
Post reply on HN