Hey Eli, I was wondering if you'd ever continue or get back to your playlist, "Introduction to Networking". That playlist peaked my interested and I've been looking forward to more of those lectures. Hope you'll get back to that sometime in the near future 🙂👍🏻
i dont speak english very well but your english is very well understandable, thanks for intro to GPT API. You did pause between words and working with intonation, may be this is the core why :)
Your video inspired me to to try to connect to the API and I was able to get some code working. I am an Adobe ColdFusion developer so I asked ChatGPT to provide ColdFusion code to connect to ChatGPT API. What it gave me worked with a few minor edits. The possibilities are endless with this technology. Thanks for providing these videos. Very helpful!
dude i've been grinding on your videos. this one is great, thanks for posting. excellent info about the copyright considerations, will play into a decision we're currently making about branding for our selves and our clients
Copyright does not necessarily go to the technician or “the person who hit the button”. So the monkey story does not completely explain it. If the producer of a movie hires a director and a cameraman, the cameraman does not get to override the rights; see what I mean. Also, (and ai am a professional photographer) I know famous srtists in several examples who no longer hold the camera or brush. The creation is in the prompt, idea or inspiration and the interest for the idea to be made into some artifact. If the ai takes my prompt, it is using my power not the other way. It is just a technical tool carrying out my wishes and intent. The second I put my hand on the artifact, it is mine. Then, in law ai can say “if anyone contests ownership please come forward”. Even if an image is similar to anothers, if it is different by even a small percentage then it is original and mine. Keep in mind, in court it all comes down to how rigorous you explain your position. So even if there are cases that lost, I know I personally could win any case, and that is basically the real stance on authoring an artifact and being confident it is yours.
Thank you, great content Eli as always! Is there any example of companies using this on their websites? Other than slanting some responses, I don't see how it's much better than just plugging questions into chatgpt itself. Is there a good example of a use-case with a website doing this?
Just a note, great work Eli. Question, in the future do you think in the future the AI infrastructure will be localized, akin to have a business might have it own Exchange servers today?
About tokens... I actually asked ChatGPT what exactly is a token, and it gave me a vague answer... ( by vague I mean scientifically accurate definition but hard to understand for simple folks...) So I asked it to: Give me this sentence, 1 token/line. Chat GPT's answer: Give me this sentence, 1 token/line. I also asked to give me C++ code 1 token/line, and I can't really post code in comment, but basically every opening and closing bracket, every word, every operator, every semicolomn at the end is a separate token... In conclusion it can generate a lot more text then code in low level programming languages. Python may be better in that regard... In addition to that as far as I understand the limit of 4000 tokens must contain the question as well as the previous conversation for context, because it does not have a memory, and it does not store the text... Whatever you put in, that's all the context it can work with... That combined with GPT4's verbosity is a nightmare you gonna pay for if you're using it a lot!
The crazy thing is, you don't even have to know code at this basic level! You can pseudo code in ChatGPT. For the bias example at 01:00:00, all I did was type in the following in GPT-4 and got almost the exact same response and formatted the output as a table: Create a list {slant}: slant = [christian, scientist, pastafarian, republican, democrat] Create a {question}: question = how did the world begin For slant (x), submit "How do " + slant(x)+ "s" + " think about" + question Your output should be in the format of a table with 2 columns. The first column should have rows that contain each slant(x) and the second column should have rows that contain the response to "How do " + slant(x)+ "s" + " think about" + question
And then as a follow up, you can tell GPT-4 to "Turn this into python code and put it in a window so I can easily copy it" and it will generate the code output for you!🤯
Hey man, still doing good work I see. Nothing to see here. Just the resurrected ''big tuna'' coming to check in. It's been a while. Glad to see your still at it an evolving your brand. Talk soon maybe? --The Tuna
There's a recent development with running these kinds of models locally. Especially with LlaMA, llama.cpp, Alpaca, etc... Would you ever do a seminar on this?
Eli, what if I use ai to generate me a solid image, borderline realistic, and then I use software to change, manipulate, and alter to my liking? Or what if I use multiple ai generated photos and stitch them together? Eventually the content would be unrecognizable
Why do i need a seminar if i can use chatgpt to explain to me whats relevant and i can save the info for later. Gonna be a mission to get a transcript generated of this.
Ill probably editing this comment as I watch. Q: i take it you'd most likely add something like DAN prompt to beginning of query (not sure if they work on 3.5 model) if you want to get rid of those disclaimers on queries?
I'm guessing it has something to do with his weird ass pivot to making edgy political commentary videos when it was like, dawg we have hundreds of thousands of those already and every single one is so annoying.(i forgot what he leaned but fairly certain i disagreed w/most of what he saidlmao) So yes pls get back to teaching us tech//how to pass certs roots
... the "weird assed pivot" to what you people watch, what TH-cam recommends, and what the advertisers pay well for...??? just... stop... virtue... signaling... PLEASE... ... I'm focused on Silicon Dojo. Doing tech content is pretty much sure to kill my channel and so I'm simply recording presentations that I already created for in person events. You all won't watch them, TH-cam won't recommend them, and advertisers don't care... I'm simply squeezing this cat nipple for the last few pennies that it's worth... And to the old content, it's just not useful in "conversation" culture when I actually have to talk with CEO's and such have zero interest in trying to explain the perversions that you all are so fascinated by...
When a drone takes a photo or movie on automatic by using way points etc technically the footage was not taken by a man. Is that a coy right loop hole 🕳? #yourthoughts
I came for the tech run through and left with this overwhelming feeling of ... like you said .... SPAM is going to be way out of hand beyond what it already is. Not to mention, for every great use of AI there will be thousands of times the amount of pure junk content being created and pushed
Ok, first I know this isn't valid legal advice and to seek a proper professional if it matters but I might as well ask and maybe you can give some half decent idea at what the answer would be or point to some place that they've already been answered by someone with equivalent or more knowledge than yourself: Say you're creating a comic book series and want to copyright it Could you create a "mini" comic book manually first and copyright it, securing rights to merchandise for any characters within, then use AI to create more indepth sequels to draw more customers? Or would the fact that AI was used for those make anything within free-game for others? If it does actually give you arguable copyright then you'd obviously need to make sure anything _new_ that you wanted to merchandise was first created by a human... Which brings up a second point. If you use AI to create something, no copyright, but then re-create it yourself (or pay someone and get them to sign over their rights), can you then copyright that? Or is no copyright ever possible because AI was the first to create it? Also, how much of a modification is necessary to call something "new" and re-copyright-able. If an AI generates some amazing image of a person with a blue hat and you use photoshop to change that hat to green, does that make it a new image? Or if you take that person and throw it into a photoshopped background scene? In short, ok, so you can't copyright AI images, but how much does it take to _legally_ claim that an image originally produced by an AI is now a human work of art similar to how it's possible to claim human-produced derivatives of other human-creators are new works that don't fall under the original's copyright.
Hey Eli, I was wondering if you'd ever continue or get back to your playlist, "Introduction to Networking". That playlist peaked my interested and I've been looking forward to more of those lectures. Hope you'll get back to that sometime in the near future 🙂👍🏻
piqued
@@crashdavis4123 Thanks :)
i dont speak english very well but your english is very well understandable, thanks for intro to GPT API. You did pause between words and working with intonation, may be this is the core why :)
Your video inspired me to to try to connect to the API and I was able to get some code working. I am an Adobe ColdFusion developer so I asked ChatGPT to provide ColdFusion code to connect to ChatGPT API. What it gave me worked with a few minor edits. The possibilities are endless with this technology. Thanks for providing these videos. Very helpful!
OMG cold fusion!
The ancient language of the great ones. Kudos.
dude i've been grinding on your videos. this one is great, thanks for posting. excellent info about the copyright considerations, will play into a decision we're currently making about branding for our selves and our clients
i KNOW... i COLULD SEE THIS anywhere... but I choose ELI.
Eli making the current educational material we really need and that nobody else is!
Thank you.. I need this information so that I am somewhat educated as these changes take place.. 😊
Copyright does not necessarily go to the technician or “the person who hit the button”. So the monkey story does not completely explain it. If the producer of a movie hires a director and a cameraman, the cameraman does not get to override the rights; see what I mean. Also, (and ai am a professional photographer) I know famous srtists in several examples who no longer hold the camera or brush. The creation is in the prompt, idea or inspiration and the interest for the idea to be made into some artifact. If the ai takes my prompt, it is using my power not the other way. It is just a technical tool carrying out my wishes and intent. The second I put my hand on the artifact, it is mine. Then, in law ai can say “if anyone contests ownership please come forward”. Even if an image is similar to anothers, if it is different by even a small percentage then it is original and mine.
Keep in mind, in court it all comes down to how rigorous you explain your position. So even if there are cases that lost, I know I personally could win any case, and that is basically the real stance on authoring an artifact and being confident it is yours.
Eli the ChatGPT guy
Thank you, great content Eli as always!
Is there any example of companies using this on their websites?
Other than slanting some responses, I don't see how it's much better than just plugging questions into chatgpt itself.
Is there a good example of a use-case with a website doing this?
Just a note, great work Eli. Question, in the future do you think in the future the AI infrastructure will be localized, akin to have a business might have it own Exchange servers today?
Love to see Eli back
Eli making tech content is a good feeling.
About tokens... I actually asked ChatGPT what exactly is a token, and it gave me a vague answer... ( by vague I mean scientifically accurate definition but hard to understand for simple folks...)
So I asked it to: Give me this sentence, 1 token/line.
Chat GPT's answer:
Give
me
this
sentence,
1
token/line.
I also asked to give me C++ code 1 token/line, and I can't really post code in comment, but basically every opening and closing bracket, every word, every operator, every semicolomn at the end is a separate token...
In conclusion it can generate a lot more text then code in low level programming languages. Python may be better in that regard...
In addition to that as far as I understand the limit of 4000 tokens must contain the question as well as the previous conversation for context, because it does not have a memory, and it does not store the text... Whatever you put in, that's all the context it can work with... That combined with GPT4's verbosity is a nightmare you gonna pay for if you're using it a lot!
Oh, Eli makes Computer videos again, nice to see.
I hope the view counter reaches Louis Rossmann numbers soon.
Good to see you after looong time Eli!!!!
The crazy thing is, you don't even have to know code at this basic level! You can pseudo code in ChatGPT. For the bias example at 01:00:00, all I did was type in the following in GPT-4 and got almost the exact same response and formatted the output as a table:
Create a list {slant}: slant = [christian, scientist, pastafarian, republican, democrat]
Create a {question}: question = how did the world begin
For slant (x), submit "How do " + slant(x)+ "s" + " think about" + question
Your output should be in the format of a table with 2 columns. The first column should have rows that contain each slant(x) and the second column should have rows that contain the response to "How do " + slant(x)+ "s" + " think about" + question
And then as a follow up, you can tell GPT-4 to "Turn this into python code and put it in a window so I can easily copy it" and it will generate the code output for you!🤯
Silicon Dojo!
Hey man, still doing good work I see. Nothing to see here. Just the resurrected ''big tuna'' coming to check in. It's been a while. Glad to see your still at it an evolving your brand. Talk soon maybe? --The Tuna
Will you be starting the Silicone Dojo Bank soon 🤑
Excellent introduction !
Thank you 🙏
There's a recent development with running these kinds of models locally. Especially with LlaMA, llama.cpp, Alpaca, etc... Would you ever do a seminar on this?
Great Host , Thanks for share video
I have been seeing this guy video since I started using computers, I think you should start building short content.
Eli the half marathon guy
interesting subject man
Eli, what if I use ai to generate me a solid image, borderline realistic, and then I use software to change, manipulate, and alter to my liking? Or what if I use multiple ai generated photos and stitch them together? Eventually the content would be unrecognizable
Why do i need a seminar if i can use chatgpt to explain to me whats relevant and i can save the info for later. Gonna be a mission to get a transcript generated of this.
Thanks
very good video, very interesting.
surely you can add some ADS and make it to run for free
Ill probably editing this comment as I watch.
Q: i take it you'd most likely add something like DAN prompt to beginning of query (not sure if they work on 3.5 model) if you want to get rid of those disclaimers on queries?
Why did you remove all your videos??
Eli said awhile ago that he doesn’t keep his videos up, because it hurt the type of brand he wanted to display of himself
I'm guessing it has something to do with his weird ass pivot to making edgy political commentary videos when it was like, dawg we have hundreds of thousands of those already and every single one is so annoying.(i forgot what he leaned but fairly certain i disagreed w/most of what he saidlmao)
So yes pls get back to teaching us tech//how to pass certs roots
... the "weird assed pivot" to what you people watch, what TH-cam recommends, and what the advertisers pay well for...???
just... stop... virtue... signaling... PLEASE... ...
I'm focused on Silicon Dojo. Doing tech content is pretty much sure to kill my channel and so I'm simply recording presentations that I already created for in person events. You all won't watch them, TH-cam won't recommend them, and advertisers don't care... I'm simply squeezing this cat nipple for the last few pennies that it's worth...
And to the old content, it's just not useful in "conversation" culture when I actually have to talk with CEO's and such have zero interest in trying to explain the perversions that you all are so fascinated by...
Sometimes if you have the original link, so bookmarked it, way back machine may have a copy, yes it also has TH-cam content on it.
@@elithecomputerguy Fwiw, your political series was very entertaining Eli. I really enjoyed that.
Why is Most in the comments section so sore..🤣
Anyways, keep up the work
Hey Eli! You still doing sum strange for a lil bit of change?
Do LangChain next! It's written in python and would be a perfect addition =]
When you said, all the training on Django is trash,Yes, I ve bought every book, made easy and only you know how much Django tutorials, suck!!?:-)
When a drone takes a photo or movie on automatic by using way points etc technically the footage was not taken by a man. Is that a coy right loop hole 🕳? #yourthoughts
I came for the tech run through and left with this overwhelming feeling of ... like you said .... SPAM is going to be way out of hand beyond what it already is. Not to mention, for every great use of AI there will be thousands of times the amount of pure junk content being created and pushed
Ok, first I know this isn't valid legal advice and to seek a proper professional if it matters but I might as well ask and maybe you can give some half decent idea at what the answer would be or point to some place that they've already been answered by someone with equivalent or more knowledge than yourself:
Say you're creating a comic book series and want to copyright it
Could you create a "mini" comic book manually first and copyright it, securing rights to merchandise for any characters within, then use AI to create more indepth sequels to draw more customers? Or would the fact that AI was used for those make anything within free-game for others? If it does actually give you arguable copyright then you'd obviously need to make sure anything _new_ that you wanted to merchandise was first created by a human...
Which brings up a second point. If you use AI to create something, no copyright, but then re-create it yourself (or pay someone and get them to sign over their rights), can you then copyright that? Or is no copyright ever possible because AI was the first to create it?
Also, how much of a modification is necessary to call something "new" and re-copyright-able. If an AI generates some amazing image of a person with a blue hat and you use photoshop to change that hat to green, does that make it a new image? Or if you take that person and throw it into a photoshopped background scene?
In short, ok, so you can't copyright AI images, but how much does it take to _legally_ claim that an image originally produced by an AI is now a human work of art similar to how it's possible to claim human-produced derivatives of other human-creators are new works that don't fall under the original's copyright.
1:06:00 Does there exists a Centrists between these Republicans and Democrats?
I DEMAND NEW SOCIETAL COMMENTARY COMPUTER MAN!
Warning Will Robinson! #stayepic
Noice
Is Silicon Dojo dead?
Eli FTW.
1:10:32 WTF HAHAHAHAHAHA
AI should replace Eli so that I can stop ignoring him.
Don't worry, someone will make sure the key is expired.
Not bias against republicans considering what potential elected rebublican's say in front of microphones. Sorry!
User is a slur 😂
wtf is this
I don't know man, we just woke up here. 🤣
Every time I see this guy always remind me of ed-ther-net.
Thank you for this Eli 🫡