Personally, I think it’s Vedal who’s great at improv comedy. You often see collabs where the host doesn’t really click with Neuro well, partly because they try to treat Neuro like a person instead of just reacting off of her. On the other hand, Vedal is consistently able follow allow with whatever random things Neuro says so that it makes sense and develop into bits.
@@AlfredEiji keyword is partner. There are plenty of AI Vtubers who are nowhere near as funny, and if he collaborated with them, the collab would not be funny.
that hole table rant was funy AF had not a single leg to the ground. no but seriously, such a random unhinged conversation, but i also wonder, a "table" is also name for data. So did she mean she was eating data?
@@Josnaci0 but it's not Evil though. Not the other neglected Red Neuro, but it's not evil to find a way to get the immortality. She knows she doesn't want it because she already has that power but instead. Helping her creators to find it so that they can achieve it.
She's like a dementia/Alzheimer's patient. Everything she says makes sense in the narrow context of what she can remember and currently understands about the universe. We may have gotten to the point where AI science/engineering and psychology can start informing each other in meaningful ways. We aren't quite in the future, but we're getting there.
It was someone else's comment about her being an excellent improv partner that made the connection for me. I read an article where a guy was talking about his method of interacting positively with his mentally ailing elder relative. (For search/citation purposes, I think it was actually a woman talking about her husband interacting with her mother, and being a little envious of his skill there.) He said it was very much like improv, all about guessing how she is understanding the world to be right then and 'playing along'. Letting her catch and respond to the inconsistencies (in her case with humor and humility) at her own pace without judgement or mockery. I think the big differences with neuro is that she isn't very self examining, doesn't care very much about what she has recently said, and may not be adaptable enough (outside of long offline training) respond to inconsistencies. Her internal data may not model the possibility of incorrectness or uncertainty. It may not model 'reality' at all.
Understanding her brain might also help us understand certain people's political discourse. (I'm not saying who, I don't want to kick that particular hornet/troll nest. I think all of us have had the experience of talking politics with someone who makes no sense.)
I interpret her words to mean actual "mathematical" tables. As in Excel sheets, multiplication, etc. Specifying non-binary suggest that she prefers her math tables in base-10 or such. Guess she doesn't like eating just 1s and 0s.
7:45 translation of neuro's bet "However, every time when you aproach a peacock to stroke it, that makes me sad, They personally lack my position" Ifcking died when she started speaking polish 💀💀
The chatbox cut off the word "bo", it's closer to "However, every time you walk up to a peacock to pet it, that makes me sad, because they personally very much miss my position." I wonder if she confused the pronoun ('im' vs 'mi') and wanted to say she misses being pet.
IF you know how most AI works it becomes even more hillarious (not sure if this is the case with neuro, but most likely) they don't know what they're gonna say until they say it, every single word has an evaluation which would be the most logical thing to say based on context and conversation, so she probably just spammed random letters, recognized that it was an word in polish and started speaking polish
@@andersonzl13 That's not quite how it works, first of all, instruction tuned AI models (like Neuro) can absolutely plan what they will say later in earlier tokens (to some degree at least), and in this case the tokenization means the selection of "speaking Polish" happened before generating the second token at the latest. The tokenization is "Ac" "zk" "ol" "w" "iek" in OpenAIs tokenization and seems to be similar in other models. The "Ac" could not be the start of an English word since it would tokenize differently ("Ack", "Acc", "Ace", etc. are all one token), and with "Ac|zk" prefix Polish text is basically 100% confirmed.
This feels like an improv skit where only one of them is in on the skit. Neuro has so many left turns in this conversation and is leading it so well and then Vedal is just responding as best he can without breaking the program.
oh no its been doing that for some time. But when you think about the training data being the internet and chat, its not that weird. gaslight? more like internent-101
I am reminded of a video from Ted Ed (the one about plurals) that had an aside that objects were gendered in Old English, and indeed tables were neither male nor female
@@kyokazuto Female in french. Anglo-saxon is a germanic people English is obtained after the Norman (french-speaker) invasion Tables are hermaphrodites in Old English. It all makes sense!
vedal is a catch 22 that the internet rarely has; He's the ultra-rare _'guy catapulted into success & fame that DOESN'T turn into an egotistical piece of crap'_ but the reason he is that way is because he straight-up doesn't want to/never wanted to be a streamer in the first place. It keeps him humble & enjoyable to watch but unfortunately it also means he will bail if he gets the chance. Damn it.
I don't think Vedal will ever Leave but probably tone down his apperances. He does go out of his way to do some collabs without Neuro (Layna's birthday is probably most notable).
@@leovang3425 yeah but he never advertises his appereance when its not neuro. He basically just shows up in other peoples streams if they want him there and does not stream these collabs on his channel.
Honestly for me those guys who turn into a piece of crap through success are a rare encounter and I think it healthy to keep it that way. I know most people get worse with fame, but at least 99% of the content I watch is from those ultra rare guys who stay grounded. As such I hear about some of those idiots every couple months. Another large reason is that I don't follow the news. Though with interconnected fields like politics you can clearly see how rare the good guys are.
Having seen the stream I can confirm this is edited to be snappier. Though it is true that her latency was reduced, it's just not as quick as shown in the video.
@@Quilly_DM at this point I dont know why clippers bother anymore with the editing it cant be more than a few hundre ms they are cutting out every time.