[ad_1]
Artificial Intelligence is everywhere these days, with technologies like ChatGPT and DALL-E encroaching on human artistic creativity and expression. Sometimes it yields interestingly absurdist results, such as that AI-generated Seinfeld-like show Nothing, Forever. Other times, it feigns humanity, like the AI-controlled VTuber Neuro-sama, which has recently moved onto reacting to content in front of tens of thousands of Twitch chatters—and, according to its creator, has plans to work with “human streamers” in the future. Things are getting even weirder on the internet.
Neuro-sama may look like a typical VTuber, but the popular livestreamer is actually operated by artificial intelligence. When pressed for details via Twitter DM on how Neuro-sama learns and communicates, her creator, Vedal, simply said that “she uses a combination of state of the art AI models and algorithms. Her chat AI for example is powered by Large Language Models.” And, according to their Twitch channel, the VTuber “mostly” uses the programming language C#. The AI has gotten in trouble in the past for denying the Holocaust and saying women’s rights don’t exist, but Vedal has taken steps to prevent that from happening again.
“I’d like to reiterate that the only control I have over what Neuro-sama says on stream (outside of singing) is the ability to cancel anything that could potentially compromise Twitch’s ToS,” Vedal told Kotaku. “Otherwise, she reacts to videos on her own, she interacts with chat on her own, and loves her viewers on her own.”
Read More: AI-Controlled VTuber Streams Games On Twitch, Denies Holocaust
Hopefully, Neuro-sama’s edgelord antics are officially over, because she’s pivoted to doing reaction content. A popular form of content creation, react videos are just what they sound like: Twitch streamers and YouTubers sitting in front of their cameras, queuing up tons of videos, and providing their raw reaction during a livestream. It’s almost voyeuristic in a way as you peek into the creator’s mind, getting to know how they feel about content and, in some ways, the world at large.
G/O Media may get a commission
![Ring Car Cam](https://i.kinja-img.com/gawker-media/image/upload/c_fill,f_auto,fl_progressive,g_center,h_264,pg_1,q_80,w_470/682bcb4fe9b324f8d3d1b7d660158972.jpg)
It’s a camera. For your car.
The Ring Car Cam’s dual-facing HD cameras capture activity in and around your car in HD detail.
Twitch watchdog Jake Lucky took to Twitter to share a snippet of Neuro-sama reacting to Swedish streamer Hans “Forsen” Fors getting bodied by Elden Ring’s Radagon of the Golden Order and Elden Beast. Lucky noted that Neuro-sama was “averaging 5k concurrent viewers” watching Forsen get his shit absolutely rocked, but the video-on-demand (VOD) tells a totally different story in terms of view counts which climb as high as 85,000 to 120,000 viewers depending on the video and its content.
“Forsen can’t play that well when he’s sober,” Neuro-sama said, repeating “lol” multiple times over. “What? His reactions are hilarious.”
This reaction is a small segment from Neuro-sama’s February 1 livestream in which the AI-controlled VTuber watches tons of content. This includes videos such as a moose shaking its antlers off, Minecraft stuff, other VTubers, and old Markiplier clips of him playing Five Nights At Freddy’s, among other things. A couple of times throughout the livestream, which pulled in over 80,000 viewers, Neuro-sama watched Nothing, Forever, a 24/7 Seinfeld-like AI-generated show made by Mismatch Media using DALL-E, OpenAI GPT-3, and more. It was like AI Inception, and I have no idea what’s going on in this timeline. Is this an alternate reality? Send help.
I was curious how Neuro-sama queued up the videos she’d react to—or if she even did it herself—but Vedal told me, “the first react stream was mostly a test stream so I was feeding videos based off of suggestions from the Discord server. The reception from the fans was very positive, and I love to see people so invested in this project. In the future I have plans to do streams where the chat can suggest videos for her to watch (after they’re checked by a moderator of course).”
Read More: AI-Generated Seinfeld-Like Twitch ‘TV Show’ Is Peak Absurdity
This is what ultimately has me worried. While it’s fascinating that AI is sophisticated enough to pose as a human Twitch streamer reacting to popular and/or interesting videos, it’s also disconcerting that artificial intelligence is capable of such things. It could render entire jobs, especially those in the content creator space, obsolete, particularly as Neuro-sama grows in popularity. With nearly 140,000 Twitch followers and roughly 42,000 YouTube subscribers—and videos on both platforms garnering tens if not hundreds of thousands of views—it may only be a matter of time before Neuro-sama overthrows all of your faves. More than that, though, it may become difficult separating a real VTuber from an AI-generated one. The lines between human and tech are beginning to officially blur, y’all.
When asked if Vedal believes Neuro-sama has the potential to supplant real steamers, he said, “I honestly don’t know. There seems to be an audience for it, and there’s room for many different forms of entertainment at the moment. There’s no reason they can’t work together, she even has some collaborations with human streamers coming up soon!”
“Collaborations with human streamers” is some seriously dystopian shit, but maybe Neuro-sama’s popularity can give some “human streamers” a boost. I can’t help but watch Neuro-sama with curious pessimism. Growing as a content creator is hard as you’re vying for attention in a competitive, imbalanced system. Monetizing your content is even harder, especially if it doesn’t generate the eyeballs necessary to make revenue.
How much money Neuro-sama makes is anyone’s guess (I asked Vedal and he declined to disclose how much revenue the streamer rakes in, suggesting it “wouldn’t be fair”), but one thing is clear: he plans to “[take] her to places never before seen by an AI.” AI streamers are here to stay. So, get used to it.
[ad_2]
Source link