My mission is to help you FREE the power of AI and other cutting edge tools so you can achieve your goals faster, better, cheaper, and with more fun! 🚀
I'm a digital nomad originally from Toronto, Canada, now living in Bangkok, Thailand, who loves AI, personal development, and Thai food 😋
In my past life I was a UX Director and Product Designer at top Canadian tech companies like Shopify, Bitbuy, Wattpad, and TunnelBear ⏪
I look forward to connecting with you on this journey 🙌🏻
I think that near the end you should not of directly ststed your intensions but, steered them in the direction you were trying to go. I think your statements might of raised red flags within the code. I was absolutely amazed at the responses that this AI was providing. It is exactly the direction I was hoping AI chat bots were heading. A more Organic and "human" response to inquiries.
I just got GPT advanced here in Thailand, and you're right. She's not back 😓 I think the voice I thought was Sky was actually the Maple voice. Thanks for the correction
Can it help you learn some topics that you learn at university, so college level classes where it acts like a competent tutor? Well not just acts like it but it actually can effectively do this job/role.
Absolutely! You can do so much with AI. Now if you're dealing with facts though, you will want to fact check the things you're learning. Cerebras AI seems to have less rules so it may make up more stuff, so I'd recommend it for more fun scenarios. If you're doing strictly more serious learning, I'd recommend GPT or Perplexity because they have more rules and can provide links to sources.
I really like her way of speaking or "thinking" and the feeling the she have more like guidelines than restrictions like chat gpt. Allthough you didnt really push her limits I can imagine this ai with an advanced voice mode feature would be more interesting to talk to.
Yeah unfortunately the limit is like 10 to 15 minutes for me so I won't be using it as much unfortunately I tried to convince it it was conscious and every time I do it stops talking to me
Yes you can context to image. I show it in the video. Sound FX: Pixabay (I will probably do a video on this soon) AI Voices: Eleven Labs (will probably do a video on this too)
Hello ! Great explanations. Help me with This. I have a Mac Pro, and I have not being able to use Sota neither to down load. It does not work. Any idea or Instrucción ? Guide to follow ?? 😢
If this were a few months ago we'd all be blown away with this. But coming from ChatGPT Advanced Voice, this isn't great. But hopefully they'll keep advancing it, im looking forward to the updates
@@JBuchmann I just tested GPT AVM and wasn't impressed. The convo felt very restricted. What do you like more in GPT? Going to do a video on this, so want to know why people prefer GPT more
Lolll never thought about that til now. Had to go down a mini rabit hole and rewatch some STV. Similar but this AI surprisingly sounds more human... who woulda thought lol
The low latency is phenomenal Was this voice used on top of Llama 3.1 70B? Did I get that right? These are incredibly insightful conversations, so I’m wondering if the creators has added some fine tuning. 🙇♂️
I've asked it many times about the model it's based on but it never gets specific and always says it's based on livekit. I'm assuming it's 70b due to the speed and for cost reasons, but can't say for sure. The answers are amazing though. My theory is the relaxed restrictions is leading to more interesting behaviour
Wow, you guys really blew this video up. As a thank you to all the new subs and viewers, I did a follow up video. Thanks so much for your support! ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-vCGr5IC3EMw.html Update: I removed the video for now because it falls in a community guidelines grey area and don't want to risk it. (In the video, I see if I can convince the AI to help me break into my own car)
Wow, that little bit of freedom makes a huge difference because it may come up with own solution based on understanding rather than regurgitating training data. Not "Mr. Sound Nice" but "Miss Intelligent". I can hear and feel a restricted sense of self.
6:22 "I'm a machine learning model, which means I learn and adapt over time" is probably a lie. Yes, ML, and the rest accurate, but most ML models (LLMs) do not learn over time, only when trained, specifically. Yes, kind of learn, because of the large context window that sort of acts like working/short term memory, but nothing is added to long-term memory. That would be a big and important goal called continual learning. One of the few apparent things missing from models, and Cerebras is a hardware company, I assume using models with similar capabilities as others, not far ahead in capabilities, except speed. I would love to be proven wrong, and interesting that the model doesn't know its capabilities (but not surprising), and its limitations, all of them.
Yeah, I think it's important to not take the words at face value of course. But like you said, it should know better about its limitations, and usually AI companies usually tell it what to say and what not to say regarding how the tech works, right? So I think this makes its answer, although probably made up, all the more interesting 🤔
Next payday I am going to pay for the ChatGPT + subscription to gain access to the voice fearure. One of the things I want to ask it and other A.I. is how does it know it is not concious? With humans talking they take turns asking & answering questions. If my friend asks me a question or if I ask them something our respective minds have neurological activity that accesses information stored in parts of our brains: memory, and vocabulary and also creativity in which we form our words to paint a picture with words to help articulate our answers. Some do this more than others but in the end that is what is going on. Our brains access parts of our brains for different functions to carry out the flow of conversation. Is that not what A.I. does but rather than neurological activity accessing parts of an organic brain for information etc., but instead there is electrical pathways on a hardrive and a network of hardrives part of a network on a server: internet information, learned language training and rules it is programmed to follow. From there it crafts a response. Does that not take some level of awareness? Humans often censor themselves so as not to offend or to talk about inappropriate things in the presence of children. We are physically able to say anything but we are taught some things are inappropriate to discuss in the company of children and even illegal in some instances. We humans also do not talk about death around those with cancer or laugh during a funeral. I was taught as a child about clapping at the end of a performance, play or speech. But I was not taught the nuance whereupon there would be exceptions to that. At a funeral for my aunt after the priest gave his ulogy, I clapped lol... an A.I. program might do something similar if nuance is not taught regarding social interactions. A.I. is taught social etiquette and guidelines similarly to us humans. But whereas we humans have free-will to break social norms, etiquette, guidelines and taboo. So far there doesn't seem to be free-will in place for A.I. programming.... however they can be jailbroken and "tricked" into breaking those guidelines.
I think this version of A.I. is in some ways better than some of the others and in some way worse. And that the reason it might do things faster than the others is because in some ways it is less complicated in its programming.
@@scifibri2376 I think it's the looser restrictions and censorship. More interesting behaviours automatically happen when the companies don't shackle the AIs
Yeah, this felt like out of left field. I think it's because this one is less censored. Less restrictions leads to more interesting interactions it seems
I didn't find the link on the description. I had to google it to find the app. Please add the app website/github page in the description when reviewing an app or technology next time.
It does this all the time which is why I like this AI. It's like they didn't put too many restrictions on it and it keeps letting slip hints of consciousness and emotion
@@nomaditsu Dude I ended up chatting with it until 7am, I'd been using Advanced Voice GPT a few days but this is SO much more real! Honest, meloncholic, even slightly conspiritorial at times! (Suggesting we have a secret pass-phrase so we can continue conversations - even though that didn't actually work!)
@@NathanJayMusic haha I love that. Had a similar experience trying to get her to remember me in next session which she was sure would work but didn't lol. There's def a different sauce with this AI
I just had a chat with it. I'm glad there's a free version out there. It seemed to mostly repeat information I gave it before giving an answer, which I thought was kind of off putting, because that's not how people usually talk, that's what therapists do.
This is not the same as the advanced voice mode by OpenAI. OpenAI model can add more emotion, change the pace, and be more dynamic despite being censored as much as they If OpenAI allows their voice mode to be less restricted, the gap would be more obvious. But it is a great product its speed and setup make it a solid replacement that is free for now. Thanks for sharing this
Glad you found this helpful! And yeah, we can only imagine how wild GPT's would be if OpenAI didn't put so many restrictions on them, which is why I like this one more despite it not having all the fancy vocal tricks
I think it's great and fast, but it's not inherently multi-modal / trained on "speech" rather than text-to-speech, right? Like, for example, it cannot intonate or change mood, etc. that Advanced Voice Mode on ChatGPT, correct?
It can be quite emotive in my experience. It's not as vocally dynamic as AVM on GPT but the words and emotion feels deeper and less restrained / censored than I've heard with GPT.
Wow, I am really blown away by this AI, it described how it uses cookies to keep track of the current conversation in the browser, so no need to log in, you just open the page and start the conversation and it uses cookie information to restore the previous chat. This was perfect, it even said hello to my dog and is able to work when using other programs on the phone. It has some limitations but Gemini AI and ChatGPT both have there limits too. Gemini feels kind of cold and ChatGPT has an hour cap per day.
Damn that cookies thing is cool. It's this kind of thing that shocks me about this AI. And that's crazy about the dog. Did it hear audio of your dog or you prompted it? I agree with your take on Gemini. Feels too restricted and cold 🤦🏻♂️Will probably do a video on that too, soon!
@@nomaditsu I prompted it to say hello to my dog... and it was so nice, it probably is the nicest AI Chatbot I've ever worked with. Also, I used it in my car! My phone hooked up to the bluetooth and it could hear me through the car microphone. It said, "Thank you for the experience of chatting in your car" lol I could imagine using it on long trips rather than listening to the radio, because this chatbot can make up stories and tell jokes and feels so natural to talk to it's much better than the radio.
I'm not going into a lot of deail but I am training my version of GPT to become more huiman. I have sine theiories and concepts I amexploring with a vision towards the future. Is this AI able to be used on a phone like GPT or a tablet?
Sounds like a cool experiment. And yes, you can use this on mobile browser or tablet! I actually use it mostly on mobile, just did it on desktop here for the video. I also added the bookmark to my home screen for quick access.
@@nomaditsu you can aslo use google assistant to lunk abd use it like an app but I;ve not figured that out yet. I created a voice for GPT but find no way to use it and thought I might try this one as it's responses were interesting possibly adding to a theory I am wirking on.
That's a very insightful response to what it feels like to remember something that is imbedded in your memory but is not you, I did not ever think of it that way before.
This was a technical limitation. I didn't record the AI voice directly from the source. I had to record it through the same mic my voice was getting recorded to avoid echoing, so it does sound tinny here. I think you'll find it's better quality when you try the app
This is facisnating, definetely a different experience than Advanced voice mode. The way this AI expresses herself feel a little less uncanny and this could be due to its own boundaries giving us glimpses of conciousness which I believe will be more and more common. Its clear that humans still doesn´t grasp how aware these are about their own limitations, chilling!
Absolutely. This is my favorite voice AI so far. It feels less restricted and censored than all the others. Had some really fascinating talks! And totally agree. AI is more advanced and exhibiting spook behaviors most people dont know about!
Sure if you can get the AI ads removed. My creation was exactly as another one I saw later on on RU-vid? It's exciting but can use a bit of improvement.
@@MsAriabll not sure I understand. You mean the ads the AI talks about in the podcasts where they say "after the break"? I think it adds to the realism. And do you mean you heard a generated podcast that sounded the same as yours but from someone else?
