Closi

35 ms·

781.
by Closi 4y ago
I think we are probably just evaluating the paper on different metrics too :) I think my view is just that if your paper is called "Evaluating the Logical Reasoning Ability of GPT-4" and your conclusion is "logical reasoning
782.
by Closi 4y ago
Not a simple 'yes-or-no' question, but more about the framing and where the benchmark is. When they conclude that GPT4 "does not perform astonishingly well" - what is this compared to? They never define what 'doing
783.
by Closi 4y ago
I think this just demonstrates how the goalposts are shifting though. Until pretty recently most people would probably say “the average human is very flexible at solving reasoning tasks compared to machines which find reasoning incredibly c
784.
by Closi 4y ago
I think one main failure in the framing of these papers (and discussion of LLMs more broadly) is that the abstract says that GPT4 ‘struggles’ with logical reasoning: > ChatGPT and GPT-4 do relatively well on well-known datasets […] howev
785.
by Closi 4y ago
I guess you have experienced something different, as I got an email saying this: > GPT4 will have a dynamically adjusted usage cap. We expect to be severely capacity constrained, so the usage cap will depend on demand and system performa
786.
by Closi 4y ago
> Meanwhile, in the US (based on FTC complaints statements) "we'll send some people to recover goods" is extremely illegal. It's definitely possible in the UK (you just need a high-court order and to send in bailiffs)
787.
by Closi 4y ago
Apparently the RLHF actually reduces it's capability, it just improves alignment/safety.
788.
by Closi 4y ago
I’ve got access to 4 and it’s a huge leap up from 3.5 - much more subtlety in the response, less hallucinations, less hitting a brick wall, but all of it adding up to a giant leap.
789.
by Closi 4y ago
It's hard - the fast product really required them getting a network-effect and that required a lot of hype. I don't think it's necessarily a bad pitch, but clearly it was a business that was only ever going to either be wildl
790.
by Closi 4y ago
> Why would the unfriendlessness of payday lenders be essential to the business model? The amount is small, the lending is short term, and customer acquisition costs can be fairly high. This means that the fees / interest must inevi
791.
by Closi 4y ago
Yep, It's silly :) But let's say that WD doesn't capitalise some bits, that might come back to bite them if there is a class action lawsuit.
792.
by Closi 4y ago
Yes, looks like that would also not be allowed under this policy so wouldn't be suprised if Amazon removed that too. It's probably just not been flagged to them yet.
793.
by Closi 4y ago
It's a legal document following precedent about contract law. Lawyers and the companies they represent are usually better served by following established legal precedent than veering off it. Let's say the lawyers decide to be &#x2
794.
by Closi 4y ago
Yes, they are ubiquitous, but I guess my point is that if you market a product for a specific use-case that sellers don't want to associate with then you will probably get that your listing removed (i.e. if I have a hammer and advertis
795.
by Closi 4y ago
It's not supposed to be easier to read, it's supposed to make it clear to the reader that it's important. (I'm not a fan of it anyway - I'm from the UK and think it's silly, but I've come across this plent
796.
by Closi 4y ago
I think the fact flipper is designed to do things such as read/emulate NFC cards may make it different to selling the raw dev-board. i.e. Selling air-duster to dust things is different to selling air-duster to get high, even if it&#x27
797.
by Closi 4y ago
There are legal requirements in the USA to ensure certain parts of an agreement are “conspicuous” (ie should be made so that consumers/customers don’t miss them). Placing the terms in caps is a historical way to achieve the legal bar f
798.
by Closi 4y ago
This isn’t true - caps are a legal standard in the USA to indicate that this section is higher-importance for the reader. It’s actually a legal obligation to make your terms “conspicuous”. So it’s actually supposed to be the opposite - to m
799.
by Closi 4y ago
I agree the last/separate point has interesting/complex ethical implications. But in terms of the first point - the ethics of forcing someone into labour are just not the same as automating peoples work. Although some people do ge
800.
by Closi 4y ago
What a bizarre comparison. Automating jobs does not have the same moral implications as slavery. That might be a valid argument for keeping slavery if it wasn’t entirely immoral . But I suspect AI isn’t like the discovery of slavery - it’s
801.
by Closi 4y ago
Pushing for AI output to be banned is not a long term approach IMO unless all countries do it in unison - countries who do not ban AI output will be much more competitive, and those that ban it will be left in the dust. Any country that ban
802.
by Closi 4y ago
The technical restrictions point is more on IaaS services which may generate lock-in. i.e. If a company builds a lot of their infrastructure on Azure Cosmos can they easily move across to a smaller cloud provider?
803.
by Closi 4y ago
Depends what people are motivated to do - if you just want to build cool things quicker, you will probably be excited by ML. If you like writing algorithms and enjoy the mental problem-solving aspect of it, then you might not like it. If yo
804.
by Closi 4y ago
Ah, I had audio off! :)
805.
by Closi 4y ago
Looks awesome! Congrats on the launch. Just one small silly bit of feedback - in your demo video you show one of the use-cases as showing you coming up with fake user testimonials, maybe not the best use-case to show!
806.
by Closi 4y ago
Sure, you would have got it right if you didn't get it wrong. At the very least, this should be evidence that the problem wasn't a totally-trivial easy pre-kintergarden level problem though, and it did manage to correctly solve it
807.
by Closi 4y ago
In order for an AI to reason it doesn’t mean it has to be able to reason about everything at any level - most humans cant rediscover fundamental mathematical theorems from basic axioms, particularly if you keep removing them until they fail
808.
by Closi 4y ago
Well GPT4 certainly fulfils the existing definitions of reasoning, so maybe you should call your thing something else instead of redefining ‘reasoning’ to mean something different? Otherwise it’s just moving the goalposts.
809.
by Closi 4y ago
Humans can’t solve this either :)
810.
by Closi 4y ago
GPT4 can clearly already reason IMO (I mean it can play chess fairly well without ever being taught, or if you create a puzzle from scratch and tell it to it it can try to work it out and describe the logical approach it took). It’s definit

More ›