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

ChatGPT Plus

openai.com

Post reply on HN
Print view View original

771–780 of 1001 posts

Re: ChatGPT Plus

#771

Post by thorawayz »

fragmede wrote: post #34
forrestthewoods wrote: post #5

Earlier quoted context omitted.

Yes please. It really needs a “safe search off” mode. It can have a big disclaimer “if you ask for something offensive then you’ll get it”.

I can just see the article now: OpenAI is run by a bunch of violent racist sexist rapists. Using the new "safe search off mode", we found out ChatGPT's underlying biases, and it turns out that it's horrible, the people that made it are horrible, and you're a horrible person for using their service. But really we're horrible for writing this article. OpenAI doesn't want that story to be written, but after Microsoft Ta…

[flagged]

Re: ChatGPT Plus

#772

Post by wraptile »

loteck wrote: post #338

Why isn't the inaccuracy of ChatGPT a more widely discussed topic on HN? Lots of excitement for responses that are often wrong when it counts. Best way I've heard it characterized is that ChatGPT is "dreaming" when providing responses. Yet it seems like users are relying on its responses to varying degrees?

Yeah, it's definitely confidently wrong at times. I got in the habbit of asking it "are you sure?" after some iffy responses and often it would correct itself.

I stumbled into a few programming questions where it be confidently wrong and refuse to correct itself and it took me as long to verify it as it would have taken to research it myself.

I think people are still surprised that it's right more often than not.

Re: ChatGPT Plus

#773

Post by thorawayz »

a5withtrrs wrote: post #719

Not being able to do simple math is a surprisingly huge bug IMO Me: 5000 * 5000 * 5000 ChatGPT: The result of 5000 x 5000 x 5000 is 125,000,000,000,000. Me: 5000 * 5000 ChatGPT: The result of 5000 x 5000 is 25,000,000. Me: 5000 * 25000000 ChatGPT: The result of 25,000,000 x 5000 is 125,000,000,000. Integrations with services like wolfram alpha would be nice as well (although shouldn't be necessary to solve basic mult…

[flagged]

Re: ChatGPT Plus

#774

Post by p-e-w »

zemo wrote: post #756
p-e-w wrote: post #731

Earlier quoted context omitted.

It's not a "bug" in the traditional sense, but a consequence of how language models work. Probability distributions over token chains are very powerful tools for generating human-like written text, but they perform poorly at symbolic inference. Humans have the same limitations, which is why math is a separate topic taught in schools that focuses not on knowledge but on algorithms. Understanding natural language synta…

if you ask it a question and it gives you a confident answer that's 100% wrong that's a terrible experience and a non-working product. "it's not a bug in the traditional sense" is technically correct but missing the point entirely.

ChatGPT is not a product, at least not for the purpose of answering arbitrary questions with a high degree of correctness. Nobody is claiming that ChatGPT is, or is supposed to be, the ultimate answer generating machine. If you are relying on the current generation of language models for that purpose you are making a huge mistake. They are a technology demonstration for what future AI products suitable for the mass market might look like, nothing less and nothing more.

Re: ChatGPT Plus

#775

Post by ynab4 »

m463 wrote: post #308
ezekg wrote: post #205

Earlier quoted context omitted.

You use it consistently during the workday and it's still not worth $20/mo?

This is one of those puzzling things to me. You own your life - why not spend your own money for the things that make you and your life better? Who cares? I worked at a job where I had a small, crappy monitor. I made decent cash. I just bought a large decent monitor and brought it into work. I ended up using it for many years. My life was significantly better. I've done that at several jobs since then, and NEVER regr…

"Pay the billion dollar companies for solving problems that they created!"

"Consoom more tech hardware made in China!"

"Don't think too hard about your monthly expenditures!"

"Trust the science!!!1!"

Well you've convinced me, buddy.

Re: ChatGPT Plus

#776

Post by int_19h »

ElevenLathe wrote: post #283
sterlind wrote: post #219

Earlier quoted context omitted.

It's going to hit so unevenly. My partner works with children at a homeless shelter, I'm an algorithm designer. I'm certain my job will be obsolete before my partner's is. It's going to automate away nearly all pure desk jobs. Starting with data entry, like you've seen, but it'll come for junior SDEs and data scientists too. Customer service, then social media/PR, then marketing, as it culls the white collar. Graphic…

This is a still from a movie set in 1959 (Billy Wilder's excellent /The Apartment/): https://onceuponascreen.files.wordpress.com/2017/08/apt.jpg Those people are all doing white collar jobs nobody does today: adding up columns of numbers with desktop adding machines, retyping manuscript memos, responding to customer and vendor mail, summarizing other hand-generated reports into newer more digestable reports (which ar…

I am not confident at all, because I don't see any law of nature or society that dictates that we shall always come up with new meaningful jobs whenever we render some of the old ones obsolete. It's true that we've done so before, but that was also when our economic development seemed exponential; surely we know better now? In any case, without a clear understanding of the mechanism, who's to say it's not just luck and/or survivor bias?

Alternatively, we can come up with meaningless jobs. Which is, of course, just UBI in disguise and with more incidental waste.

Re: ChatGPT Plus

#777

Post by jliptzin »

m463 wrote: post #308
ezekg wrote: post #205

Earlier quoted context omitted.

You use it consistently during the workday and it's still not worth $20/mo?

This is one of those puzzling things to me. You own your life - why not spend your own money for the things that make you and your life better? Who cares? I worked at a job where I had a small, crappy monitor. I made decent cash. I just bought a large decent monitor and brought it into work. I ended up using it for many years. My life was significantly better. I've done that at several jobs since then, and NEVER regr…

The question isn't really if $20 is worth it, the question is can you get 90% of the way there for much less? It sounds like he found a way, and of course he's going to do that.

Re: ChatGPT Plus

#778

Post by rhaway84773 »

extr wrote: post #107

I've been using ChatGPT pretty consistently during the workday and have found it useful for open ended programming questions, "cleaning up" rough bullet points into a coherent paragraph of text, etc. $20/month useful is questionable though, especially with all the filters. My "in between" solution has been to configure BetterTouchTool (Mac App) with a hotkey for "Transform & Replace Selection with Javascript". This i…

Have you been able to find uses for chatGPT where it’s better than the top result in Google?

I’ve consistently found that Google returns similar results. The only difference is that if it’s a longer snippet I need to click into the website, but even then, it’s usually faster than chatGPT.

Re: ChatGPT Plus

#779

Post by int_19h »

r00fus wrote: post #214
electrondood wrote: post #200

Earlier quoted context omitted.

The future is bifurcated into those who invested in AI companies in the 2020s, and those on UBI.

I strongly doubt the UBI part as it goes against the "religion" of Wall Street and most politicians.

That's not actually true - if you look around, you'll find a surprising number of people who are right-wing economically overall, but specifically support UBI; and this is especially popular in tech circles, which, if it really goes as OP described, would be the ones amassing power. Two simple reasons: first, when suffering is too obvious and blatant, it's harder to not notice it; and second, you can only push people too far before they riot, and it's cheaper to pay them just enough to avoid that.

The question, rather, will be: what kind of UBI - "just enough to prevent large-scale unrest", or "as much as we as a society can afford"? The former is not necessarily an utopia or even close to it; think The Diamond Age and The Expanse.

Re: ChatGPT Plus

#780

Post by wraptile »

MuffinFlavored wrote: post #206

I wonder how close they are to breaking even/making a profit on $20/mo per person? I know the narrative here is that Microsoft if footing/subsidizing a massive GPU Azure bill. How massive? $10,000,000/month? How many people do we think would convert (sign up and start paying $20/mo) next 30 days? 10,000? 100,000? 1,000,000?

I think some estimates were that avg query costs around 1-5 cents USD so to break even the users has to make less than 13-66 queries per day which is quite reasonable. I use it daily and from a brief glance I avg ~20 queries/day, which will probably go down once novelty wears off.

I also doubt that openai is in need of making profit now, just covering operation costs is a huge win at this stage.

Post reply on HN