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

Google just launched Bard, its answer to ChatGPT

technologyreview.com

Post reply on HN
Print view View original

21–30 of 81 posts

Re: Google just launched Bard, its answer to ChatGPT

#21

Post by marban »

kryptiskt wrote: post #14

I don't think Google will win this race, not because they lack the technical expertise, but because they are too cheap. This is so telling: "We’re releasing it initially with our lightweight model version of LaMDA. This much smaller model requires significantly less computing power, enabling us to scale to more users, allowing for more feedback" They are richer than Croesus, that is no reason for them to hold back an…

They also lack taste — Just as Steve Jobs used to mock Microsoft; but it increasingly seems that the tables are turning.

Re: Google just launched Bard, its answer to ChatGPT

#22

Post by ergl »

gorbypark wrote: post #17
bko wrote: post #5

I've been using ChatGPT a lot for various tasks and found the results very helpful. There are often errors but errors that I would make. With non programming tasks it can sometimes be word association and misstate the concept but it's almost always helpful and again the mistakes are reasonable. When I asked Bard to list the months in alphabetical order, it failed to do so. I saw this on a tweet but confirmed that it…

That's pretty surprising it can't do the months in alphabetical order. I just tried with LLaMA 30B. This is running on my MacBook Air with llama.cpp, quantized to 4 bits, and is a "raw" model (not alpaca / instruction tuned at all). Obviously some models will be better at some things (and will have different outputs given the same inputs) and worse at other, but you'd think Bard would be able to do that pretty simple…

Both answers are wrong though?

Re: Google just launched Bard, its answer to ChatGPT

#23

Post by singularity2001 »

gorbypark wrote: post #17
bko wrote: post #5

I've been using ChatGPT a lot for various tasks and found the results very helpful. There are often errors but errors that I would make. With non programming tasks it can sometimes be word association and misstate the concept but it's almost always helpful and again the mistakes are reasonable. When I asked Bard to list the months in alphabetical order, it failed to do so. I saw this on a tweet but confirmed that it…

That's pretty surprising it can't do the months in alphabetical order. I just tried with LLaMA 30B. This is running on my MacBook Air with llama.cpp, quantized to 4 bits, and is a "raw" model (not alpaca / instruction tuned at all). Obviously some models will be better at some things (and will have different outputs given the same inputs) and worse at other, but you'd think Bard would be able to do that pretty simple…

gpt4:

months in alphabetical order

Here are the months in alphabetical order:

April August December February January July June March May November October September

Re: Google just launched Bard, its answer to ChatGPT

#24

Post by Random_Person »

I've been rather unimpressed with Bard when asking it "common sense" type of information. It's limited memory is also a problem when trying to have a meaningful conversation with depth.

I spent a few hours last night trying to work on a story outline with Bard's help and while individual responses were sometimes okay, Bard regularly forgot about plot points and character traits we decided on earlier in the conversation and that was frustrating. ChatGPT is much better at conversational memory, but provided much less interesting results to individual queries.

Unlike other commenters, I think these artificial limitations aren't about spending or competency, but rather out of fear. Google is afraid of their bot turning racist, or saying things that could embarrass them.

Re: Google just launched Bard, its answer to ChatGPT

#25

Post by ancientworldnow »

glofish wrote: post #12

Bard is a better search engine. A really good one at that. But it is the same old thing - only much better. It lacks the spirit, sparkle, the pizzaz and surprising "intelligence" of ChatGPT. The service, if released before ChatGPT, would not have made the splash ChatGPT did.

I asked Bard for local restaurants and it listed ten. Two had the wrong type of food (saying it was Mexican instead of Italian for example), two were closed and had been so for years, and two were entirely made up. Only four were completely correct.

I asked it for information about a show and it summarized it correctly but then hallucinated the hosts as two entirely made up people with fake careers. It even linked websites to the show with the correct information but made up this false info.

It's even worse than Google search which is already middling.

Re: Google just launched Bard, its answer to ChatGPT

#26

Post by KeplerBoy »

agilob wrote: post #9
FooBarWidget wrote: post #6

Earlier quoted context omitted.

4 months after ChatGPT is "too little too late"? Where does this entitlement come from?

There were like at least 60 LLama and alpaca releases in the last 3 weeks! Google needs to be at least 3x better!!

No, they don't. They're still the world's most visited page [citation needed] if they start to automatically integrate Bard into every search they have their 99% of market share even with an inferior model.

They have a few more years to pull that off.

Re: Google just launched Bard, its answer to ChatGPT

#27

Post by franze »

gorbypark wrote: post #17
bko wrote: post #5

I've been using ChatGPT a lot for various tasks and found the results very helpful. There are often errors but errors that I would make. With non programming tasks it can sometimes be word association and misstate the concept but it's almost always helpful and again the mistakes are reasonable. When I asked Bard to list the months in alphabetical order, it failed to do so. I saw this on a tweet but confirmed that it…

That's pretty surprising it can't do the months in alphabetical order. I just tried with LLaMA 30B. This is running on my MacBook Air with llama.cpp, quantized to 4 bits, and is a "raw" model (not alpaca / instruction tuned at all). Obviously some models will be better at some things (and will have different outputs given the same inputs) and worse at other, but you'd think Bard would be able to do that pretty simple…

Months are not in alphabetical order.

And largest city is an ambiguous answer: per area, moskow is, by population its istanbul.

Re: Google just launched Bard, its answer to ChatGPT

#28

Post by MagicMoonlight »

kryptiskt wrote: post #14

I don't think Google will win this race, not because they lack the technical expertise, but because they are too cheap. This is so telling: "We’re releasing it initially with our lightweight model version of LaMDA. This much smaller model requires significantly less computing power, enabling us to scale to more users, allowing for more feedback" They are richer than Croesus, that is no reason for them to hold back an…

The actual product is the user interactions. ChatGPT etc is trained on examples in the thousands. What google will be doing is trying to get 100s of millions of labelled interactions by getting the users to do it for them.

Then they’ll use those to train its final model.

Re: Google just launched Bard, its answer to ChatGPT

#29

Post by enlyth »

Random_Person wrote: post #24

I've been rather unimpressed with Bard when asking it "common sense" type of information. It's limited memory is also a problem when trying to have a meaningful conversation with depth. I spent a few hours last night trying to work on a story outline with Bard's help and while individual responses were sometimes okay, Bard regularly forgot about plot points and character traits we decided on earlier in the conversati…

It already did a pretty good job of embarrassing them, social media is full of Bard failure screenshots. It couldn't even write a Javascript function to add two numbers.

Re: Google just launched Bard, its answer to ChatGPT

#30

Post by jsnell »

kryptiskt wrote: post #14

I don't think Google will win this race, not because they lack the technical expertise, but because they are too cheap. This is so telling: "We’re releasing it initially with our lightweight model version of LaMDA. This much smaller model requires significantly less computing power, enabling us to scale to more users, allowing for more feedback" They are richer than Croesus, that is no reason for them to hold back an…

You're making the assumption that quality is the only feature that matters, rather than it being a multi- dimensional tradeoff where we don't know at all what tradeoffs consumers actually prefer. For example, a smaller model will run faster, right?

At this exact moment, ChatGPT is giving me about a word every 2-3 seconds. It's basically useless. But even when the system is not overloaded, there's a noticeable delay.

How much quality would you be willing to trade off to get results instantly instead? Or the inverse, how much better would the results need to be to justify a 5x longer processing time? It seems hard to believe that ChatGPT happened to be released at exactly the optimal tradeoff. (And obviously it's also unlikely that Google launched with exactly the right tradeoff.)

Post reply on HN