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

GPT-3 has no idea what it’s talking about

technologyreview.com

Post reply on HN
Print view View original

191–200 of 323 posts

Re: GPT-3 has no idea what it’s talking about

#191

Post by ctoth »

I keep wanting to write a long explanation of just why this is so... silly? to read? But Gwern has already done the hard work. [0]

The only other bit I'd like to mention is that GPT-3 uses exactly none of the new techniques that have been coming out in the last two years that would have significant impact on text generation. From working methods to apply GANs to text, to far more efficient transformer models that can handle longer sequences. For instance [1] [2] [3] for better direction, or [4] [5] [6] for efficiency.

Or perhaps the outside view might help. After seeing GPT-2 last year, did you expect GPT-3 would work as well as it does after just naively scaling up the number of parameters with nothing else?

[0https://www.gwern.net/newsletter/2020/05#gpt-3

[1 ] http://arxiv.org/abs/1905.09922

[2] https://github.com/anonymous1100/D_Improves_G_without_Updati...

[3] http://arxiv.org/abs/2006.04643

[4] http://arxiv.org/abs/2007.14062

[5] http://arxiv.org/abs/2006.04768

[6] http://arxiv.org/abs/2002.05645

Re: GPT-3 has no idea what it’s talking about

#193

Post by visarga »

tasogare wrote: post #143
throwawaygh wrote: post #113

Some of the criticism in this comment section is completely fair — the authors are providing exactly the type of prompts that GPT-3 breaks down on and some of these examples might be cherry-picked continuations. And the authors do have personal interests at stake. (NB, the exact same criticism is true about a lot of articles lauding GPT-3, which is why public discussion of GPT-3 in general is such a dumpster fire.) S…

> Several other researchers I know — very good researchers who happen to have been publicly critical of GPT-2 — have not been given access. Wow, this is an incredible nasty move. This is allow telling about the confidence they have in their model.

GPT-3 is what it is no matter who got invited to the beta. I have personally tried it and it clearly is above any language model up to date, and made me seriously consider prompt programming as an emerging field.

What I would like to see next is: to extend the corpus with task specific data where available, making it more performant on these specific tasks; to have a searchable memory module (ability to retrieve text from its corpus or new text additions); to enlarge its context window; to enlarge the corpus, including more non-english content; to be trained on more than just text: images, audio and video, for grounding. It also needs a leash - a second model to stop it from generating profanity or insulting things and better ability to control the expected size and format of the generated text: dialogue, article, novel, etc.

Re: GPT-3 has no idea what it’s talking about

#194

Post by monkpit »

junon wrote: post #183
bigyikes wrote: post #99

Earlier quoted context omitted.

Lacking “understanding” doesn’t make GPT-3 less impressive and also doesn’t make comparisons to human abilities unwarranted. I read the prompt, and I expected that this was the beginning of some kind of fiction. In my mind, it sounded like I was reading the beginning of a somebody’s dream. What does it even mean to understand something? Because naively, it looks very much like GPT-3 and I have a shared understanding…

It's not about impressiveness - surely, it's impressive. However, the article is more or less critiquing the discourse surrounding the model - namely, that there is a strange misconception floating around that it's somehow a general purpose AI that can understand and think about the world similar to a human. Which, of course, it cannot. If the claims about GPT-3 were accurate, there'd be a lot less of a flare-up abou…

I fail to see where openai is making any false claims

Re: GPT-3 has no idea what it’s talking about

#195

Post by sillysaurusx »

throwawaygh wrote: post #178
lumost wrote: post #153

Earlier quoted context omitted.

For OpenAI to become a healthy and profitable business, GPT-3 will require them to generate ~50-300 million dollars from the model. This could realistically only occur if they cost-effectively fine-tune away the more egregious problems in beta - or convince enough investors that their next model with a 100 million dollar price tag will be able to handle something approximating AGI for realistic applications. This is…

~50-300 million sounds like an over-estimate, and the investment in GPT-3 is miniscule compared to SDC. Plus, GPT-3 generates value even without direct product impact. I'm sure MSFT sales reps are already folding tons of nonsense about gpt-3 into their Azure pitches. Industry labs like OpenAI and DeepMind replace/augment the "Research & Development" model with a "Research & Marketing" model.

Seems spot-on. One trick to estimate a startup's burn rate is to multiply their number of employees by $200k. It's not too accurate, but it's within the ballpark.

So how many employees does OpenAI have? Supposing they have 500, that's a burn rate of $100M/yr.

250 employees, $50M/yr.

100 employees, $20M/yr.

Re: GPT-3 has no idea what it’s talking about

#196

Post by mrfusion »

jfk13 wrote: post #107
bko wrote: post #79

Earlier quoted context omitted.

I stopped reading right after that clothes comment to comment exactly what you had. If you even provide the simplest context of question answer gpt3 answers reasonably [Prompt] Q: What is the day after Tuesday? A: Wednesday Q: Yesterday I dropped my clothes off at the dry cleaner’s and I have yet to pick them up. Where are my clothes? A: [gpt3] A: They are in the dryer. Another give away that the article wouldn't be…

> Q: Yesterday I dropped my clothes off at the dry cleaner’s and I have yet to pick them up. Where are my clothes? > A: [gpt3] A: They are in the dryer. Sorry, but I don't think this can be considered "reasonable". There's a huge difference between a dry cleaner's and a dryer. Which nicely illustrates, I think, just how little GPT3 "knows" what it's talking about.

This seems like the kind of error a small child could make. Doesn’t completely understand what you’re talking about but understands the question and throws out its best guess.

Re: GPT-3 has no idea what it’s talking about

#197

Post by monkpit »

rirarobo wrote: post #180
gweinberg wrote: post #132

Earlier quoted context omitted.

To be fair to the AI, stirring lemonade with a cigarette is so batshit insane that there really can't be a sensible continuation.

Hmm a sensible continuation to an absurd situation? Sounds like fun fiction. > At the party, I poured myself a glass of lemonade, but it turned out to be too sour, so I added a little sugar. I didn’t see a spoon handy, so I stirred it with a cigarette. But that turned out to be a bad idea because it promptly dissolved into my drink, creating a most unpleasant concoction, with an aroma which evoked memories of my gran…

Is your argument that a human can write better prompted fiction than gpt3?

Re: GPT-3 has no idea what it’s talking about

#199

Post by mcguire »

hnuser123456 wrote: post #184
throwawaygh wrote: post #171

Earlier quoted context omitted.

> I do think there's an argument to be made that we should try to keep it in a good light to further interest in the field. How many more articles do we need like the one linked by OP? This is how winter happens and is a decidedly anti-scientific mindset. Do you consult with non-technical/low-technical execs on AI investments? If you did, the answer to your question would be obvious. For every dollar Microsoft and VC…

Is a winter not a widespread lack of interest in the field? I would stay step 0 is getting people interested by showing what's possible, and step 1 is showing what's not currently possible but should be.

A widespread lack of interest caused by a failure of a previous technology in the field to live up to it's hyped claims.

Re: GPT-3 has no idea what it’s talking about

#200

Post by nmca »

sjg007 wrote: post #94
skatesor wrote: post #32

Earlier quoted context omitted.

You have to remember the AI cannot produce sentences or even words that someone else didn't already write. I'd totally agree it is 'aware' if it could meaningfully come to conclusions like these without getting them from someone else. You might say "don't all humans learn things from someone else" which is not really true because at some point there had to be a first person who learned something completely independen…

Words it has to have seen,, but sentences no. It is generating new sentences. I think it can “name” things as well.

This is not true, BPE can back off to characters so arbitrarily character strings can be generated.
Post reply on HN