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

Comparing Google and ChatGPT

twitter.com

Post reply on HN
Print view View original

181–190 of 280 posts

Re: Comparing Google and ChatGPT

#181

Post by tedunangst »

margalabargala wrote: post #140
houstonn wrote: post #133

Quoted post unavailable.

> Biology is not confused about this. Males produce gametes (sperm) and females produce large gametes (ova). There are no intermediate gametes, which is why there is no spectrum of sex. Biological sex in humans is a binary system. This is not entirely true, due to the existence of various kinds of intersex genotypes, which may produce no gametes, or both gametes (functionality notwithstanding). Biological sex in huma…

GPT seems to "understand" certain terms as synonyms, like binary and bimodal.

Re: Comparing Google and ChatGPT

#182

Post by jsnell »

spaceman_2020 wrote: post #139
hncel wrote: post #135

I work at Alphabet and I recently went to an internal tech talk about deploying large language models like this at Google. As a disclaimer I'll first note that this is not my area of expertise, I just attended the tech talk because it sounded interesting. Large language models like GPT are one of the biggest areas of active ML research at Google, and there's a ton of pretty obvious applications for how they can be us…

The problem for Google isn't just technical, it's organizational. The entire organization and all its products are built around ads. If a new product comes along that drastically reduces the number of pages a user views, what happens to the ad revenue? Right now, every click, every query is an impression. But if there's an all-knowing AI answering all my questions accurately, what incentive do I, as a user, have to s…

> Right now, every click, every query is an impression.

The vast majority (maybe 95%?) of my searches don't show ads.

Re: Comparing Google and ChatGPT

#183

Post by magicalist »

bushbaba wrote: post #103
impulser_ wrote: post #88

Earlier quoted context omitted.

You are providing only a small part of the result. If you provide the full result it make prefect sense why Google would suggest it. When you search "when did the moon explode?". The full result is actually > About an hour after sunset on June 18, 1178, the Moon exploded. That's what it looked like to five terrified, awestruck monks watching the skies over the abbey at Canterbury, in southeastern England, anyway. Whi…

Doesn’t seem that crazy for a future iteration of GPT to offer sources.

That's a pretty different type of query and storage you'd need, though.

Re: Comparing Google and ChatGPT

#184

Post by thewarrior »

s17n wrote: post #176
thewarrior wrote: post #171

Earlier quoted context omitted.

Why should serving only this model cost as much as all of Google ?

GPT3 costs something like 5 cents a query. At 20 dollars a month, that would be 400 queries a month. I don't know about you but I'm pretty sure I do at least an order of magnitude more Google searches than that.

I don’t think every query needs this

Re: Comparing Google and ChatGPT

#185

Post by j2kun »

hn_throwaway_99 wrote: post #128

Google is only "done" if you never actually use it to find sites on the web. For nearly all of these examples I was thinking "WHY DON'T YOU JUST TRY CLICKING THE TOP LINK?" E.g. the first link for writing differential equations in LaTeX, I thought the first result, https://priyankacool10.wordpress.com/2013/10/15/writing-diff... , provided excellent, helpful examples. That is, if anything, I'd be quite satisfied with…

The LaTeX examples given by ChatGPT are also... strange? Nobody would write the same differential equation twice with two different notations right after each other (it uses y' and dy/dx in two stacked aligned equations)

Re: Comparing Google and ChatGPT

#187

Post by crummy »

29athrowaway wrote: post #122
CommieBobDole wrote: post #61

Earlier quoted context omitted.

I ran across a site a while back which just seems to be common questions fed to GPT-3; the answers all make perfect grammatical sense, but they're also hilariously wrong. A bunch of middle school kids are probably going to get an F on their papers and simultaneously learn something about the importance of verifying information found on the internet. https://knologist.com/has-any-rover-landed-on-venus/ "The average ca…

Just wait a couple of years. You are not thinking fourth-dimensionally.

Like, once half the text that GPT7 is trained on was generated by GPT1-6?

Re: Comparing Google and ChatGPT

#188

Post by macintux »

rafaelero wrote: post #169
macintux wrote: post #158

Earlier quoted context omitted.

Humans will generally either provide a confidence level in their answers, or if they’re consistently wrong, you’ll learn to disregard them. If a computer is right every time you’ve asked a question, then gives you the wrong answer in an emergency like a grease fire, it’s hard to have a defense against that. If you were asking your best friend, you’d have some sense of how accurate they tend to be, and they’d probably…

If the AI is correct 90% of the time, you can be reasonably sure it will be correct next time. That's a rational expectation. If you are at a high stake situation, then even a 1% rate of false positive is too high and you should definitely apply some verifications. Again, I don't see the danger.

Ultimately I think the danger is that the AI sounds like it knows what it’s talking about. It’s very authoritative. Anyone who presents content at that level of detail with that level of confidence will be convincing.

You can hear doubt when a presenter isn’t certain of an answer. You can see the body language. None of that is present with an AI.

And most people don’t know/care enough to do their own research (or won’t know where to find a more reliable source, or won’t have the background to evaluate the source).

Re: Comparing Google and ChatGPT

#189

Post by Simon321 »

tfsh wrote: post #11

These are addressing two very different concerns but framed as a singular one. Google is first and foremost a search engine - it searches the web for answers, the key point being the answers need to exist on the web. The other is a machine learning model tasked with deriving answers, and sometimes - if not very often answers will be provided in an authoritative tone whilst being completely and utterly incorrect. Goog…

Nevermind lamda, they have Flan-U-PaLM which is even better. They have the best LLM know to man. But it's private.

Re: Comparing Google and ChatGPT

#190

Post by jasonjmcghee »

jsnell wrote: post #182
spaceman_2020 wrote: post #139

Earlier quoted context omitted.

The problem for Google isn't just technical, it's organizational. The entire organization and all its products are built around ads. If a new product comes along that drastically reduces the number of pages a user views, what happens to the ad revenue? Right now, every click, every query is an impression. But if there's an all-knowing AI answering all my questions accurately, what incentive do I, as a user, have to s…

> Right now, every click, every query is an impression. The vast majority (maybe 95%?) of my searches don't show ads.

Wow. Very different from my experience. If I'm not using Brave, often the first 5ish results are ads.
Post reply on HN