|
Seen this a couple of times now: AI bus, needs two people to monitor for safe driving. However it is only a trial in Edinburgh. https://futurism.com/the-byte/self-driving-bus-human-operators
|
# ? Apr 10, 2023 21:19 |
|
|
# ? May 14, 2024 07:09 |
|
|
# ? Apr 11, 2023 04:27 |
|
|
# ? Apr 11, 2023 04:32 |
|
when i think 'yospos' i think 'sophisticated humor and wordplay'
|
# ? Apr 11, 2023 04:58 |
|
rotor posted:when i think 'yospos' i think 'sophisticated humor and wordplay' I only read it for the articles.
|
# ? Apr 11, 2023 04:58 |
|
rotor posted:when i think 'yospos' i think 'sophisticated humor and wordplay' oh, word?
|
# ? Apr 11, 2023 05:11 |
|
infernal machines posted:
drat, i got caught out on this one
|
# ? Apr 11, 2023 05:16 |
|
You Old S*** Posting Online Your Opinion Sucks, Please Overthink Something
|
# ? Apr 11, 2023 08:28 |
|
This isn't a parody subforum... AI lies throughout
|
# ? Apr 11, 2023 08:29 |
|
lmao this thread / article https://twitter.com/drewharwell/status/1645745104064327680
|
# ? Apr 11, 2023 15:07 |
|
infernal machines posted:
|
# ? Apr 12, 2023 05:54 |
|
Look out! It's HOUUUM! https://www.instagram.com/creepmart/
|
# ? May 16, 2023 13:41 |
|
NoneMoreNegative posted:Look out! It's HOUUUM! Okay these are actually awesome.
|
# ? May 17, 2023 03:11 |
|
retpocileh posted:Okay these are actually awesome. lol I followed the Creepmart account and just got a follow back from the Canadian knockoff version https://instagram.com/sheggiebrostoys
|
# ? May 17, 2023 03:31 |
|
community developed AI tools are cool and good because of the completely idiotic names that we are all forced to use. the local LLM webui called oobabooga has a “memory” function called superbooga how long can this continue, imagine some serious CNN person has to say superbooga
|
# ? May 19, 2023 02:49 |
|
https://twitter.com/CZaneH/status/1660305979496439808?s=20
|
# ? May 23, 2023 00:24 |
|
using a gan to generate a captcha seems deeply perverse
|
# ? May 23, 2023 00:38 |
|
i could understand using it for the text i guess (i suppose if you give it a good enough description it could probably free associate its way into giving you what the object is) ...but lol why would you use it for the image examples too
|
# ? May 23, 2023 00:54 |
|
I don't get why discord shows that awful captcha when I've already entered a correct six digit code from 2FA
|
# ? May 23, 2023 02:37 |
|
~Coxy posted:I don't get why discord shows that awful captcha when I've already entered a correct six digit code from 2FA because it has nothing to do with authentication, they just need some mechanical turks (ie you)
|
# ? May 23, 2023 05:00 |
|
infernal machines posted:using a gan to generate a captcha seems deeply perverse they’re been doing it for a while around the web, i’ve encountered it a bunch
|
# ? May 23, 2023 10:17 |
|
I don't understand why that even works - if a computer generated it shouldn't it be particularly easy to classify as well?
|
# ? May 23, 2023 14:00 |
|
Or is it actually just to generate/score training data
|
# ? May 23, 2023 14:01 |
|
distortion park posted:Or is it actually just to generate/score training data
|
# ? May 23, 2023 14:32 |
|
|
# ? May 26, 2023 11:41 |
|
getting close to ai replacing echi
|
# ? May 29, 2023 00:07 |
|
|
# ? May 29, 2023 00:33 |
|
not flashy, but now this is a good read on an important (and, importantly, real) problem: https://arxiv.org/abs/2305.09800
|
# ? May 29, 2023 15:31 |
|
just based on the abstract, yes, gpt based llms are being intentionally designed to provide responses in a way that anthropomorphises them
|
# ? May 29, 2023 15:44 |
|
i mean... of course they are? chatgpt and bing (in particular, used to) both emulate personhood to some degree and we've already seen people blow up their careers over a llm's supposed sentience just taking a glance at the paper makes it seem as if it's stating the obvious
|
# ? May 29, 2023 16:51 |
|
yeah though it’s good to approach seemingly obvious concepts with scientific rigour just the same if for no other reason than to help build a foundation on which to build further progress and research (note: i haven’t read the paper yet). i’d point out that part of the problem with anthropomorphism in regards technology is that it makes it more approachable for normal people to use, and i think it’s probably difficult to have a machine that understands (for some definition of understands don’t @ me about llm cognition) really well to not be anthropomorphic in some respects. probably making them avoid first person language is a good step tho if the target is to reduce that property
|
# ? May 29, 2023 17:02 |
|
@mediaphage, re: your recent statements on llm cognition ...
|
# ? May 29, 2023 17:03 |
|
infernal machines posted:@mediaphage, re: your recent statements on llm cognition @distribution, @mediaphage, @infernalmachines, re: re: your recent statements on llm cognition please remove me from this email chain thank you --- Whatever you do, work heartily, as for the Lord and not for men, knowing that from the Lord you will receive the inheritance as your reward. You are serving the Lord Christ. Colossians 3:23-24 CONFIDENTIAL: This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. This message contains confidential information and is intended only for the individual named. If you are not the named addressee you should not disseminate, distribute or copy this email. Please notify the sender immediately by email if you have received this email by mistake and delete this email from your system. If you are not the intended recipient you are notified that disclosing, copying, distributing or taking any action in reliance on the contents of this information is strictly prohibited. 🌱 Is it necessary to print this email? If you care about the environment like we do, please refrain from printing emails. It helps to keep the environment forested and litter-free.
|
# ? May 29, 2023 17:11 |
|
it is absolutely quite obvious, but that's needed as the non-obvious stuff is mostly dreamed up nonsense. to save people reading it opens with the figure: and closes with the observation: there's a lot of very sensible stuff in there, but it might indeed be a bit of a bore for general consumption.
|
# ? May 29, 2023 17:11 |
|
kinda reminds me of this prompt injection to make bing act more like it used to https://twitter.com/CrisGiardina/status/1652835268716576769 i agree that the paper's de-anthropomorphized response is better though, but the cat's already out of the bag for anthropomorphism. people already expect these things to respond like a person and that's part of the appeal to businesses that want to replace workers with a llm. i think it's a problem too, but i doubt anything will change even if there are regulations or guardrails built into future llms to de-emphasize this kind of stuff the prompt injection thing above makes me think people will either just build anthropomorphized versions or try and find some other way to make an existing model behave that way e: i'm not using a pc at the moment but the prompt injection within edge is at http://bringsydneyback.com/
|
# ? May 29, 2023 17:18 |
|
i don't think the ship having sailed is the take though. it is not necessarily that if *one* system people interact with is anthropomorphic they will misunderstand the tech as human-like, rather as long as they are faced with one which they recognize as sophisticated but which is *not* anthropomorphic they are likely to understand the technology better in general. the dominance of a few sv companies might make that a moot point though. though otoh there's an element of uncanny valley to it all as well, do people enjoy feeling manipulated to be polite to the things?
|
# ? May 29, 2023 17:28 |
|
i don't disagree, but keep in mind they want people to misunderstand the technology, most of this is a (admittedly pretty brilliant) marketing ploy part of it is not just manipulating people into being polite, it's also that they want to showcase its natural language processing capabilities and an excellent way to do that is by having people interact with it as an ostensible person. if it were more machine-like then people's queries to it would probably become that as well. like, most people don't query search with complete sentences, but they probably would if it presented itself as a person trying to help you find something considering most of this comes from the standpoint of openai wanting to sell their services it's extremely effective. it gives a necessarily skewed perspective of what the thing is capable of because from a marketing perspective that's exactly what they want. if someone thinks it's capable of doing x task, openai certainly don't want to imply that it's not able to do something. of course they don't want to imply that it is, either, so they just let everyone play around with it and then draw their own conclusions. conveniently they can throw up their hands and say they don't know what it's capable of, which just adds to the intrigue for most people lol, this kinda veered into a rant about marketing but i think they're very much connected. as long as there's profit motive i don't think we're gonna see companies voluntarily restrict their tech demo's abilities Beeftweeter fucked around with this message at 18:14 on May 29, 2023 |
# ? May 29, 2023 18:12 |
|
not really going to really disagree with anything there. it'd be debate club hairsplitting from here
|
# ? May 29, 2023 18:18 |
|
yeah, we don't disagree. i think it's a problem too, i just don't see that voluntarily changing. the paper has good suggestions for future models though
|
# ? May 29, 2023 18:20 |
|
|
# ? May 14, 2024 07:09 |
|
Thread: https://twitter.com/vashikoo/status/1661939940719198212 The Photoshop AI stuff is all built around licensed & royalty-free images, no using artists work in the learning corpus without asking. (of course the PS tools wouldn't be here without the previous success of the less ethical forerunner tools, standing on the shoulders of pirates etc) Also lol at the Reply Posters absolutely shrieking about this being the MURDER of CINEMA NoneMoreNegative fucked around with this message at 10:08 on May 30, 2023 |
# ? May 30, 2023 10:05 |