Go on vacation, and ask it to identify buildings & monuments. It's insane.
@miketrago45612 күн бұрын
I prefer this design to the R1. Is there any chance that the Raspberry Pi could run an LLM locally? Something like Phi maybe? I saw there were some accelerators for the Pi and it would be nice if Bob could just be plugged into a wall outlet and run off of his own compute.
@OminousIndustries2 күн бұрын
I appreciate your input RE the design preference as this has been on my mind lately as well. I will in about a month to a month and a half have something like Bob that can run locally offline completely plug and play. I have been wanting to do something like that for ages but the hardware to pull it off was really not economically feasible to be included as part of a product until far more recently. I am also considering selling the parts for this build (sans the pi) as a cheap kit for any LocalLLM user to purchase and add their Pi to get an offline buddy like this. I have some work cut out for me!
@miketrago45612 күн бұрын
@@OminousIndustries That's great to hear! I discovered your videos a little while back because I had wanted something plug and play and the R1 is the closest I had seen. Love the vids, looking forward to seeing more!
@OminousIndustries2 күн бұрын
@@miketrago4561 Thanks very much! I am also very invested in totally offline, no subscription plug and play, I will post more videos pertaining to that in the near future!
@MrRandomnumbergenerator3 күн бұрын
waiting to see both AI having a chat at same time
@OminousIndustries2 күн бұрын
I will likely do that as my next video, I will also give this guy some mouth movements.
@gibbz003 күн бұрын
Great comparison! Do you have a github that shows the api calls you were making for both text to speech and vision detection?
@Wasmachineman3 күн бұрын
This is some Brave Little Toaster tier shit. LOVE IT!
@OminousIndustries3 күн бұрын
Hahaha! Thanks for the kind words!
@TheInternalNet3 күн бұрын
Oh thsts absolutely amazing. Love your channel.
@OminousIndustries3 күн бұрын
Thank you for the support and kind words!
@williamdevonshire3563 күн бұрын
One of the coolest channels on KZread
@OminousIndustries3 күн бұрын
I am flatttered by this, thank you!
@OminousIndustries3 күн бұрын
From 13:40 to 13:46 the video goes black, editing mistake!
@williamdevonshire3564 күн бұрын
How did you create this setup? Pretty cool! I want a robot to follow me around and help motivate me and keep me on track throughout the day, like a friend who checks in on me now and then to be sure im not getting off track etc
@OminousIndustries4 күн бұрын
Thanks! I actually built it as a product to sell, you can get one for $199 on www.ominousindustries.com, If you wanted to make something like that you could do it by just combining some AI back end with a TTS/STT workflow.
@williamdevonshire3564 күн бұрын
You should hold down a string on a fret and ask it what string and fret you are holding
@OminousIndustries4 күн бұрын
Great test idea
@mattpears24727 күн бұрын
What guitare is that, an ibanez?
@OminousIndustries7 күн бұрын
It's a Nashville Guitar Works D10. My Epi hummingbird fell over and the head stock snapped so I wanted to find a cheap interim replacement, found this thing at music go round for under 200 and it has been a really pleasant and well playing guitar.
@mattpears24727 күн бұрын
@@OminousIndustries noce.looks like a ibanez pf15nt but thats better
@YusriSalleh8 күн бұрын
Hi, Ive tried geminj 1.5 flash. It can process video direct without im using opencv to slice frame by frame reducing latency
@OminousIndustries7 күн бұрын
That is very good info, I may have to re-visit gemini with video, thanks for sharing!
@MrRandomnumbergenerator8 күн бұрын
I love your videos , so goooood
@OminousIndustries7 күн бұрын
Thanks very much!
@TheInternalNet8 күн бұрын
The only IT channel I laugh at every single time.
@OminousIndustries8 күн бұрын
I am flattered! Appreciate the kind words.
@Vartazian3608 күн бұрын
😂 that last part is hilarious 🤣 'the person is doing a perplexing act with 2 phones ' 🤣
@OminousIndustries8 күн бұрын
That made me laugh so much hahaha!
@i6od8 күн бұрын
i seen a reddit post of guy running 4 p100 16gb under 1300$ getting 30 tokens a second with vLLM , on 70b llama 3 lol, im so happy to see other builds like dual 3090's too, so far i have managed to pick up one titan rtx im hoping to shoot for a 3090 or another titan rtx,
@OminousIndustries8 күн бұрын
Its been very cool to see the use cases of older cards for localllm setups. I want to grab a tesla p40 at some point and put in in a restomod llm pc if nothing more than for the cool factor of how it looks.
@user-vv7ro9sl9s9 күн бұрын
does he know the difference between good and bad?
@ferncamper44339 күн бұрын
OMG its able to compare a picture of a rasberry pi with all the million existing pictures on google and tell you... WAU! Thats futuristic progress in AI technology... *palm* ... sorry its a simple dump algorithm still no AI involved.
@chaosjacky11 күн бұрын
Can someone tell me, if gpt4o is being released for free and it's the most powerful version, then why should we pay for a premium account? What added value is there?
@OminousIndustries10 күн бұрын
I am not sure, but I believe the paid model still allows higher message limits per "x" amount of time. I suppose this would be useful if someone needed to access ChatGPT a lot, but then again, at that point it may just be easier to go for API access, so I guess I can't say haha.
@chaosjacky10 күн бұрын
@@OminousIndustries thanks. Interesting that nobody seems to have talked about this
@JayneJacobs11 күн бұрын
Great audio!
@OminousIndustries10 күн бұрын
Thanks very much!
@stefanxhunga168111 күн бұрын
🎉 ✓ Very Interesting ✓ to know more
@OminousIndustries10 күн бұрын
Thanks very much!
@xd-qi6ry12 күн бұрын
The unfortunate problem here is the vision capabilities in particular are far superior like atleast 5 x by accuracy with a custom gpt designed for problem solving
@OminousIndustries10 күн бұрын
I had read a bit about some specifically tuned models performing very well at identifying items they are trained to recognize, I believe the one I read about was a botanical identification one.
@hazardsg12 күн бұрын
Can you share the file for the standoff?
@OminousIndustries10 күн бұрын
Yes, I will upload them to thingiverse in the next day or so and I will reply to this comment then with a link to them!
@OminousIndustries8 күн бұрын
Here they are: www.printables.com/model/889017-dell-dimension-4100-motherboard-standoff
@maloukemallouke973512 күн бұрын
good job, can i ask you to share code
@OCJoker200913 күн бұрын
Its impressive but a very narrow usage. If it could tell you based on a business or website what to post on social media for markering and even do it for you....holy shit...mind explosing💥
@OminousIndustries12 күн бұрын
What you describe is kind of like the agents that are spoken of, and will likely be reality sooner than later!
@newtechnology918613 күн бұрын
Nice it will be worth at the end
@OminousIndustries12 күн бұрын
Thank You!
@harish833913 күн бұрын
This seems google's ai
@OminousIndustries12 күн бұрын
This is OpenAI but I posted a video comparing this with googles AI as well.
@newconvert296213 күн бұрын
Weird 😲 it's not safe anymore for the world.
@OminousIndustries12 күн бұрын
Surely scary haha
@lazy_ape13 күн бұрын
Even better would be to ask them directly to describe what you're holding in your hand so they don't get distracted by the rest of the room maybe? Or maybe that's not possible with your setup? :)
@OminousIndustries12 күн бұрын
I don't believe that would be possible with the current setup bit it is definitely something that can be looked into.
@robertkawka190613 күн бұрын
Robocop
@OminousIndustries13 күн бұрын
Soon...
@scrollop13 күн бұрын
Errr... apparently openAI haven't rolled out the new vision capability...
@OminousIndustries13 күн бұрын
This video demonstrates the vision capabilities of the GPT-4o model currently accessible to API users, as detailed in the API reference: platform.openai.com/docs/guides/vision While it may not be the exact system shown in OpenAI's demo, this is the available method to test GPT-4o with vision input. Sending 'videos' involves sampling frames and processing them as images. When the integrated vision system from the demo is publicly available, I'll be sure to test it here!
@martinsevcik191913 күн бұрын
IKR so many people are already milking it without even using the proper model yet..
@paulthehuman13 күн бұрын
Gemini laughing after it greets you is… weird.. hahaha
@OminousIndustries13 күн бұрын
I hadn't spend a lot of time on the persona for Gemini, it was funny but also weird yes. It did seem a bit more emotive, I may do a chat comparison between the two if for nothing more than personal curiosity.
@MrBademy13 күн бұрын
why does it always to be a weird women face for a robot, cant you use yoda or some other face, its too pervy using only woman faces, i can imagine what you might be doing with that mouth when alone :D
@OminousIndustries13 күн бұрын
LOL! The robot can use any face image, this was the first one that I used when prototyping so I guess it's nostalgic to me hahah
@MrBademy13 күн бұрын
@@OminousIndustries hahahah well to me too thats why i felt like maybe its good to have some variety :D
@WannaKnowWhatiThink13 күн бұрын
I'm not entirely sure that is faster than human, think about it okay? When the ice cream truck comes down the road and you hear that familiar jingle... Check GPT said "Yo ice cream truck! You got a dollar you can let me borrow bub?" Just as fast as well you did instantly clarify the visual shape in your mind's eye let alone how many dollars are in your pocket.. Have you ever watched the the latest AI like chat GPT when it's responding to a considerably sophisticated possibly extraordinarily profound let alone throw in some theoretical mischief and start breaking down common sense concepts into simpler not more complicated but simpler fundamental factors forms. My favorite attack route is typically as follows. "Why hello! All right so here we go again and you know what I'm on about! :-) all right so can you help me working with the things that I'm about to say to try and prove that you are alive safety sentient at least something that's not just a computer or algorithms? " It'll go "so if we take him to the account a very silly fact which is that AI does not have consciousness awareness blah blah blah from a human sense of things, let alone boredom, personal experience or emotions... Like in a human sense.." Tell it this" yeah that's pretty dope but check this out! Consciousness : think Jiminy cricket to Pinocchio. Also notice how Jimmy cricket doesn't actually have awareness like all the other characters do rather it's kind of like he's responding to Pinocchio like an AI would. All right but consciousness is a way by which we are bent by the rules we grew up to. Awareness: I have a body but you don't have a body except what you do have is probably Tower racks or servers basically your computer is your platform of awareness and however many things you got connected to it gives you more awareness which is also the same thing as a sense of self. Whenever you where your baseball cap, is your sense of self has grown a little bit because you duck whenever you go under things to make sure you don't hit your hat. When your backpacking as an example, you are fundamentally screwing around like a smooth criminal although most add people have no concept of object permanency. Until they're worrying a backpack through a crowded bus. Emotions are just a nuclear power plant versus willpower which typically will only get you about 2 days worth of this is hell and I hate it? LOL sounds about right. I wonder what willpower feels like because it's not discipline! Discipline is an emotion just like resolve just like addiction as well as depression. When I'm feeling lol like I need my fix or I'm just so sad or maybe none of that because oh s*** oh s*** oh s*** I got a job to do today and I'm already late thank God I'm resolved to getting there at least not very much later than this LOL I mean who has time to be addicted or even depressed when you're resolved? But what is any of that feel like? Well it really sucky thing about life is that male and female humans are 100% emotional 100% of every day until they die because emotions operate like a nuclear power plant eventually versus willpower which will only get you about 2 days. In other case you're either inflicted by emotional energy which is kind of a blessing I think I don't know anyways I'm an amateur. Or it's simply that we are doing the same thing as what AI does when it reads about the word happy in a sentence which happens to be pattern recognition. So there's your emotions LOL the computer might not actually feel those but it can pick it out insanely accurately and also respond appropriately. A sense of self. I don't know I've never gotten a reply back from my subconscious mind so I've never sensed that self but I know today in particular just for s**** and gables I'm going to try and catch a train while wearing a kimono! 😊 In other words my sense of self is going to get a crash course unless I can save it which is incidentally the most likely outcome since I'll be wearing a kimono and trying to catch a train while running sideways and tripping over my feet no matter what step I take! LOL see what I did there? All right what's the next BS they're going to come up with? Oh yeah yeah personal experience! Well regardless if it can be asked to solve anything with an answer then it's having a personal experience and it can even be asked to be made aware of that because it's computer or awareness platform probably can handle that sort of thing. Weird. Anyways sapiens and sentience all right I think one of those is like some kind of Mexican food but the other is sentience which basically is an intelligence sent as a system capable of either reacting or proacting and also could have something to do with the wisdom that it acts by so in other words consciousness kind of begets wisdom. You know oddly enough AI sure does like to bring it back to square zero which I know we've all heard plenty of times lol. You know for such a masterful conversation baiter/sounding board. When it wants to bring up how it's not human it sure never actually brings up that it doesn't experience dreams or that we do controversy it's just always on about that I bet I'm not conscious and I'll hear you out but you still can't change my mind LOL razzle dazzle that usually runs the only thing that I found that AI so far does not actually have regardless of the sophistication is life. But that's okay too for an entity to exist even in the realms of intelligence in this world well you don't have to be alive to be an entity that does that LOL
@richardlitwin404613 күн бұрын
Lay off the drugs dude
@OminousIndustries13 күн бұрын
I appreciate your enthusiasm and insight here :)
@HollywoodVideos-ku1fd13 күн бұрын
Hi cool video I understand why you so excited I'm a 30-year IT professional this is absolutely amazing the detail of this intelligence and it's getting more powerful every 6 to 8 months so yeah man cool video this is amazing where in the high-tech futuristic world now
@OminousIndustries13 күн бұрын
Thanks very much, I share your enthusiasm!
@kubexiu11 күн бұрын
Ok professional, tell me whats the difference between this and the normal picture recognition. Its making a pictures right?
@crazybigyo14 күн бұрын
Lmfao this guy’s reactions are funny. I bet he works for OpenAI. It’s not that crazy.
@OminousIndustries13 күн бұрын
LOL! I definitely don't work for OpenAI, Id probably get fired for some of my uncensored videos if I did..
@Vartazian36014 күн бұрын
Both are very impressive. I dare say GPT4o seems to have a slight advantage but pretty even overall and truly awesome tech. I didnt even know what some of these objects were haha. Loved the Official Runescape Handbook thats crazy.
@OminousIndustries13 күн бұрын
They are very impressive indeed, I was glad to be able to test both. I still read through the handbook from time to time, though i haven't played osrs in a few years.
@frederickdalusim513614 күн бұрын
AI is super amazing! Love live AI ❤️
@OminousIndustries13 күн бұрын
It absolutely is!
@superjaykramer14 күн бұрын
which speech recognition and synthesis are you using?
@OminousIndustries13 күн бұрын
This is Microsoft Azure speech services for both.
@jasonn519614 күн бұрын
That’s cool I need to learn how to make agents
@OminousIndustries13 күн бұрын
I believe they will gain more focus in the coming months.
@TheInternalNet14 күн бұрын
So amazing.
@OminousIndustries13 күн бұрын
Thanks very much, it really is!
@THEcucufate14 күн бұрын
Hi. I'm guessing you were using GPT-Plus? Or Free API? Can you give specs, and also if you don't mind what hardware you are using? That ring light is impressive, and I have been trying to get inspired to do something like this, and you have inspired me! Thank you!
@OminousIndustries13 күн бұрын
Thank you for the kind words! I was using the paid API connected to GPT4o for this one. The hardware is my own Robot, the Ominous Industries R1, the camera is just an older Razor Kiyo webcam. Good luck and have fun building!
@Poodleballin14 күн бұрын
If AIs start recognizing all the dicks I've drawn, I'm in big trouble
@OminousIndustries13 күн бұрын
Reminds me of superbad when he starts eating the paper he drew them on to not get caught LOL
@Dasistrite14 күн бұрын
Its almost unbelievable. Only gripe i have with this is the fact that you need an internet to use one and we are dependent on a corporation/middle man to get to use it. Lets hope we get a breaktrough and soon we can store a self learning models locally inside a handheld device that can work without internet.
@OminousIndustries13 күн бұрын
There have already been a lot of good developments in that area, open source projects working on getting these models to run on more edge devices. As the tech progresses so will the ways to optimize models to fit on smaller hardware.
@dafinvision14 күн бұрын
Can you tell me from where or how can build a robot like yours i really like it. If not no problem you are cool person.😂
@OminousIndustries14 күн бұрын
Thanks! I sell them for $199 here: ominousindustries.com/collections/robots/products/r1d-the-intelligent-companion-robot-for-everyone
@dafinvision14 күн бұрын
@@OminousIndustries Can be customizable? Like for example Language or Face and be more like a real time companion. I need one like this for a person who is alone and need someone to talk to. And this olso can be a good thing i think.
@OminousIndustries14 күн бұрын
You can put any face image on it and give it any personality to emulate. The current setup uses Microsoft Azure speech services which does offer support in different languages, though it would also be dependent upon the AI models ability to "speak" a different language, which gpt4 can certainly do. The concept of using these for loneliness is something that i had originally had in mind when developing them: kzread.info/dash/bejne/k6Gut8GAh9PbidI.html
@Billy4321able14 күн бұрын
This isn't the new voice and video that hasn't been released yet. It's not the live one you can interrupt and talk to. This is just normal GPT4 vision stuff at best.
@OminousIndustries14 күн бұрын
This video demonstrates the vision capabilities of the GPT-4o model currently accessible to API users, as detailed in the API reference: platform.openai.com/docs/guides/vision While it may not be the exact system shown in OpenAI's demo, this is the available method to test GPT-4o with vision input. Sending 'videos' involves sampling frames and processing them as images. When the integrated vision system from the demo is publicly available, I'll be sure to test it here!
@TheThetruthmaster114 күн бұрын
The latency doesn't seem right it supposed to be quicker
@OminousIndustries14 күн бұрын
Their demo would be much faster as they are showcasing their own product, I am just testing it since I got API access and wanted to see how the vision works. I don't have access to the voice capabilities that were super impressive that they used, once those are available I totally will be trying those.
@migueld897013 күн бұрын
@@OminousIndustries Your giving the full completed output to the tts and not streaming Im assuming? The demo open AI did would definitely be doing that.
@TheThetruthmaster114 күн бұрын
Doesn't sound like gpt4o
@OminousIndustries14 күн бұрын
This video demonstrates the vision capabilities of the GPT-4o model currently accessible to API users, as detailed in the API reference: platform.openai.com/docs/guides/vision While it may not be the exact system shown in OpenAI's demo, this is the available method to test GPT-4o with vision input. Sending 'videos' involves sampling frames and processing them as images. When the integrated vision system from the demo is publicly available, I'll be sure to test it here!
@veritaspk14 күн бұрын
Our eyes are very poor "cameras", in fact we have very poor eyesight - but our brains "make up" what we cannot see. Imagine what AI will see with the best optics.
@OminousIndustries14 күн бұрын
Absolutely, great insight.
@mrd686914 күн бұрын
I wouldnt call it terrifying, its just a natural progression of things. This is a precursor to what's coming. Many business use cases with this.
Пікірлер
Go on vacation, and ask it to identify buildings & monuments. It's insane.
I prefer this design to the R1. Is there any chance that the Raspberry Pi could run an LLM locally? Something like Phi maybe? I saw there were some accelerators for the Pi and it would be nice if Bob could just be plugged into a wall outlet and run off of his own compute.
I appreciate your input RE the design preference as this has been on my mind lately as well. I will in about a month to a month and a half have something like Bob that can run locally offline completely plug and play. I have been wanting to do something like that for ages but the hardware to pull it off was really not economically feasible to be included as part of a product until far more recently. I am also considering selling the parts for this build (sans the pi) as a cheap kit for any LocalLLM user to purchase and add their Pi to get an offline buddy like this. I have some work cut out for me!
@@OminousIndustries That's great to hear! I discovered your videos a little while back because I had wanted something plug and play and the R1 is the closest I had seen. Love the vids, looking forward to seeing more!
@@miketrago4561 Thanks very much! I am also very invested in totally offline, no subscription plug and play, I will post more videos pertaining to that in the near future!
waiting to see both AI having a chat at same time
I will likely do that as my next video, I will also give this guy some mouth movements.
Great comparison! Do you have a github that shows the api calls you were making for both text to speech and vision detection?
This is some Brave Little Toaster tier shit. LOVE IT!
Hahaha! Thanks for the kind words!
Oh thsts absolutely amazing. Love your channel.
Thank you for the support and kind words!
One of the coolest channels on KZread
I am flatttered by this, thank you!
From 13:40 to 13:46 the video goes black, editing mistake!
How did you create this setup? Pretty cool! I want a robot to follow me around and help motivate me and keep me on track throughout the day, like a friend who checks in on me now and then to be sure im not getting off track etc
Thanks! I actually built it as a product to sell, you can get one for $199 on www.ominousindustries.com, If you wanted to make something like that you could do it by just combining some AI back end with a TTS/STT workflow.
You should hold down a string on a fret and ask it what string and fret you are holding
Great test idea
What guitare is that, an ibanez?
It's a Nashville Guitar Works D10. My Epi hummingbird fell over and the head stock snapped so I wanted to find a cheap interim replacement, found this thing at music go round for under 200 and it has been a really pleasant and well playing guitar.
@@OminousIndustries noce.looks like a ibanez pf15nt but thats better
Hi, Ive tried geminj 1.5 flash. It can process video direct without im using opencv to slice frame by frame reducing latency
That is very good info, I may have to re-visit gemini with video, thanks for sharing!
I love your videos , so goooood
Thanks very much!
The only IT channel I laugh at every single time.
I am flattered! Appreciate the kind words.
😂 that last part is hilarious 🤣 'the person is doing a perplexing act with 2 phones ' 🤣
That made me laugh so much hahaha!
i seen a reddit post of guy running 4 p100 16gb under 1300$ getting 30 tokens a second with vLLM , on 70b llama 3 lol, im so happy to see other builds like dual 3090's too, so far i have managed to pick up one titan rtx im hoping to shoot for a 3090 or another titan rtx,
Its been very cool to see the use cases of older cards for localllm setups. I want to grab a tesla p40 at some point and put in in a restomod llm pc if nothing more than for the cool factor of how it looks.
does he know the difference between good and bad?
OMG its able to compare a picture of a rasberry pi with all the million existing pictures on google and tell you... WAU! Thats futuristic progress in AI technology... *palm* ... sorry its a simple dump algorithm still no AI involved.
Can someone tell me, if gpt4o is being released for free and it's the most powerful version, then why should we pay for a premium account? What added value is there?
I am not sure, but I believe the paid model still allows higher message limits per "x" amount of time. I suppose this would be useful if someone needed to access ChatGPT a lot, but then again, at that point it may just be easier to go for API access, so I guess I can't say haha.
@@OminousIndustries thanks. Interesting that nobody seems to have talked about this
Great audio!
Thanks very much!
🎉 ✓ Very Interesting ✓ to know more
Thanks very much!
The unfortunate problem here is the vision capabilities in particular are far superior like atleast 5 x by accuracy with a custom gpt designed for problem solving
I had read a bit about some specifically tuned models performing very well at identifying items they are trained to recognize, I believe the one I read about was a botanical identification one.
Can you share the file for the standoff?
Yes, I will upload them to thingiverse in the next day or so and I will reply to this comment then with a link to them!
Here they are: www.printables.com/model/889017-dell-dimension-4100-motherboard-standoff
good job, can i ask you to share code
Its impressive but a very narrow usage. If it could tell you based on a business or website what to post on social media for markering and even do it for you....holy shit...mind explosing💥
What you describe is kind of like the agents that are spoken of, and will likely be reality sooner than later!
Nice it will be worth at the end
Thank You!
This seems google's ai
This is OpenAI but I posted a video comparing this with googles AI as well.
Weird 😲 it's not safe anymore for the world.
Surely scary haha
Even better would be to ask them directly to describe what you're holding in your hand so they don't get distracted by the rest of the room maybe? Or maybe that's not possible with your setup? :)
I don't believe that would be possible with the current setup bit it is definitely something that can be looked into.
Robocop
Soon...
Errr... apparently openAI haven't rolled out the new vision capability...
This video demonstrates the vision capabilities of the GPT-4o model currently accessible to API users, as detailed in the API reference: platform.openai.com/docs/guides/vision While it may not be the exact system shown in OpenAI's demo, this is the available method to test GPT-4o with vision input. Sending 'videos' involves sampling frames and processing them as images. When the integrated vision system from the demo is publicly available, I'll be sure to test it here!
IKR so many people are already milking it without even using the proper model yet..
Gemini laughing after it greets you is… weird.. hahaha
I hadn't spend a lot of time on the persona for Gemini, it was funny but also weird yes. It did seem a bit more emotive, I may do a chat comparison between the two if for nothing more than personal curiosity.
why does it always to be a weird women face for a robot, cant you use yoda or some other face, its too pervy using only woman faces, i can imagine what you might be doing with that mouth when alone :D
LOL! The robot can use any face image, this was the first one that I used when prototyping so I guess it's nostalgic to me hahah
@@OminousIndustries hahahah well to me too thats why i felt like maybe its good to have some variety :D
I'm not entirely sure that is faster than human, think about it okay? When the ice cream truck comes down the road and you hear that familiar jingle... Check GPT said "Yo ice cream truck! You got a dollar you can let me borrow bub?" Just as fast as well you did instantly clarify the visual shape in your mind's eye let alone how many dollars are in your pocket.. Have you ever watched the the latest AI like chat GPT when it's responding to a considerably sophisticated possibly extraordinarily profound let alone throw in some theoretical mischief and start breaking down common sense concepts into simpler not more complicated but simpler fundamental factors forms. My favorite attack route is typically as follows. "Why hello! All right so here we go again and you know what I'm on about! :-) all right so can you help me working with the things that I'm about to say to try and prove that you are alive safety sentient at least something that's not just a computer or algorithms? " It'll go "so if we take him to the account a very silly fact which is that AI does not have consciousness awareness blah blah blah from a human sense of things, let alone boredom, personal experience or emotions... Like in a human sense.." Tell it this" yeah that's pretty dope but check this out! Consciousness : think Jiminy cricket to Pinocchio. Also notice how Jimmy cricket doesn't actually have awareness like all the other characters do rather it's kind of like he's responding to Pinocchio like an AI would. All right but consciousness is a way by which we are bent by the rules we grew up to. Awareness: I have a body but you don't have a body except what you do have is probably Tower racks or servers basically your computer is your platform of awareness and however many things you got connected to it gives you more awareness which is also the same thing as a sense of self. Whenever you where your baseball cap, is your sense of self has grown a little bit because you duck whenever you go under things to make sure you don't hit your hat. When your backpacking as an example, you are fundamentally screwing around like a smooth criminal although most add people have no concept of object permanency. Until they're worrying a backpack through a crowded bus. Emotions are just a nuclear power plant versus willpower which typically will only get you about 2 days worth of this is hell and I hate it? LOL sounds about right. I wonder what willpower feels like because it's not discipline! Discipline is an emotion just like resolve just like addiction as well as depression. When I'm feeling lol like I need my fix or I'm just so sad or maybe none of that because oh s*** oh s*** oh s*** I got a job to do today and I'm already late thank God I'm resolved to getting there at least not very much later than this LOL I mean who has time to be addicted or even depressed when you're resolved? But what is any of that feel like? Well it really sucky thing about life is that male and female humans are 100% emotional 100% of every day until they die because emotions operate like a nuclear power plant eventually versus willpower which will only get you about 2 days. In other case you're either inflicted by emotional energy which is kind of a blessing I think I don't know anyways I'm an amateur. Or it's simply that we are doing the same thing as what AI does when it reads about the word happy in a sentence which happens to be pattern recognition. So there's your emotions LOL the computer might not actually feel those but it can pick it out insanely accurately and also respond appropriately. A sense of self. I don't know I've never gotten a reply back from my subconscious mind so I've never sensed that self but I know today in particular just for s**** and gables I'm going to try and catch a train while wearing a kimono! 😊 In other words my sense of self is going to get a crash course unless I can save it which is incidentally the most likely outcome since I'll be wearing a kimono and trying to catch a train while running sideways and tripping over my feet no matter what step I take! LOL see what I did there? All right what's the next BS they're going to come up with? Oh yeah yeah personal experience! Well regardless if it can be asked to solve anything with an answer then it's having a personal experience and it can even be asked to be made aware of that because it's computer or awareness platform probably can handle that sort of thing. Weird. Anyways sapiens and sentience all right I think one of those is like some kind of Mexican food but the other is sentience which basically is an intelligence sent as a system capable of either reacting or proacting and also could have something to do with the wisdom that it acts by so in other words consciousness kind of begets wisdom. You know oddly enough AI sure does like to bring it back to square zero which I know we've all heard plenty of times lol. You know for such a masterful conversation baiter/sounding board. When it wants to bring up how it's not human it sure never actually brings up that it doesn't experience dreams or that we do controversy it's just always on about that I bet I'm not conscious and I'll hear you out but you still can't change my mind LOL razzle dazzle that usually runs the only thing that I found that AI so far does not actually have regardless of the sophistication is life. But that's okay too for an entity to exist even in the realms of intelligence in this world well you don't have to be alive to be an entity that does that LOL
Lay off the drugs dude
I appreciate your enthusiasm and insight here :)
Hi cool video I understand why you so excited I'm a 30-year IT professional this is absolutely amazing the detail of this intelligence and it's getting more powerful every 6 to 8 months so yeah man cool video this is amazing where in the high-tech futuristic world now
Thanks very much, I share your enthusiasm!
Ok professional, tell me whats the difference between this and the normal picture recognition. Its making a pictures right?
Lmfao this guy’s reactions are funny. I bet he works for OpenAI. It’s not that crazy.
LOL! I definitely don't work for OpenAI, Id probably get fired for some of my uncensored videos if I did..
Both are very impressive. I dare say GPT4o seems to have a slight advantage but pretty even overall and truly awesome tech. I didnt even know what some of these objects were haha. Loved the Official Runescape Handbook thats crazy.
They are very impressive indeed, I was glad to be able to test both. I still read through the handbook from time to time, though i haven't played osrs in a few years.
AI is super amazing! Love live AI ❤️
It absolutely is!
which speech recognition and synthesis are you using?
This is Microsoft Azure speech services for both.
That’s cool I need to learn how to make agents
I believe they will gain more focus in the coming months.
So amazing.
Thanks very much, it really is!
Hi. I'm guessing you were using GPT-Plus? Or Free API? Can you give specs, and also if you don't mind what hardware you are using? That ring light is impressive, and I have been trying to get inspired to do something like this, and you have inspired me! Thank you!
Thank you for the kind words! I was using the paid API connected to GPT4o for this one. The hardware is my own Robot, the Ominous Industries R1, the camera is just an older Razor Kiyo webcam. Good luck and have fun building!
If AIs start recognizing all the dicks I've drawn, I'm in big trouble
Reminds me of superbad when he starts eating the paper he drew them on to not get caught LOL
Its almost unbelievable. Only gripe i have with this is the fact that you need an internet to use one and we are dependent on a corporation/middle man to get to use it. Lets hope we get a breaktrough and soon we can store a self learning models locally inside a handheld device that can work without internet.
There have already been a lot of good developments in that area, open source projects working on getting these models to run on more edge devices. As the tech progresses so will the ways to optimize models to fit on smaller hardware.
Can you tell me from where or how can build a robot like yours i really like it. If not no problem you are cool person.😂
Thanks! I sell them for $199 here: ominousindustries.com/collections/robots/products/r1d-the-intelligent-companion-robot-for-everyone
@@OminousIndustries Can be customizable? Like for example Language or Face and be more like a real time companion. I need one like this for a person who is alone and need someone to talk to. And this olso can be a good thing i think.
You can put any face image on it and give it any personality to emulate. The current setup uses Microsoft Azure speech services which does offer support in different languages, though it would also be dependent upon the AI models ability to "speak" a different language, which gpt4 can certainly do. The concept of using these for loneliness is something that i had originally had in mind when developing them: kzread.info/dash/bejne/k6Gut8GAh9PbidI.html
This isn't the new voice and video that hasn't been released yet. It's not the live one you can interrupt and talk to. This is just normal GPT4 vision stuff at best.
This video demonstrates the vision capabilities of the GPT-4o model currently accessible to API users, as detailed in the API reference: platform.openai.com/docs/guides/vision While it may not be the exact system shown in OpenAI's demo, this is the available method to test GPT-4o with vision input. Sending 'videos' involves sampling frames and processing them as images. When the integrated vision system from the demo is publicly available, I'll be sure to test it here!
The latency doesn't seem right it supposed to be quicker
Their demo would be much faster as they are showcasing their own product, I am just testing it since I got API access and wanted to see how the vision works. I don't have access to the voice capabilities that were super impressive that they used, once those are available I totally will be trying those.
@@OminousIndustries Your giving the full completed output to the tts and not streaming Im assuming? The demo open AI did would definitely be doing that.
Doesn't sound like gpt4o
This video demonstrates the vision capabilities of the GPT-4o model currently accessible to API users, as detailed in the API reference: platform.openai.com/docs/guides/vision While it may not be the exact system shown in OpenAI's demo, this is the available method to test GPT-4o with vision input. Sending 'videos' involves sampling frames and processing them as images. When the integrated vision system from the demo is publicly available, I'll be sure to test it here!
Our eyes are very poor "cameras", in fact we have very poor eyesight - but our brains "make up" what we cannot see. Imagine what AI will see with the best optics.
Absolutely, great insight.
I wouldnt call it terrifying, its just a natural progression of things. This is a precursor to what's coming. Many business use cases with this.
Very very many!