Building an AI Startup in 8 Minutes with ChatGPT [No-Code Method]

in under 10 minutes we've got chat EPT to build us a fine-tuned version of itself that we can begin to start selling and building a million dollar business with entrepreneurs across the world right now are scrambling for a way to start building AI businesses however finding a place to start is not always easy if you have clicked on this video you're likely an entrepreneur who knows a life-changing opportunity when you see one the AI Gold Rush is upon us well you're in the right place because in this video I'm going to be telling you exactly the system I've created to help entrepreneurs build AI businesses with multiple millions of dollars with chat GPT in under 10 minutes so please don't go anywhere I'm going to be showing you the step-by-step process that you can create an AI business to make your Millions within this gold rush in just 10 minutes with chat GPT I've spent hours preparing this video for you guys to learn the exact strategy that I'm using to create AI startups in just 10 minutes using chat GPT so you do not want to go anywhere don't skip this video watch to the very end because there's some serious value being dropped for entrepreneurs out there looking to make money let's get started but before we get started my name is Liam Motley I've been doing online business since 2018 I'm a self-made serial online entrepreneur I've traveled all over the world and I'm now based in Dubai on this channel I focus on bringing you AI focused content for entrepreneurs like me who are looking to get started and get building businesses in the AI Gold Rush so if that kind of content interests you hit down below subscribe to the channel drop a like if you haven't already it really helps my channel and if you have any questions be sure to comment down below and I'll be answering those too ah you're an entrepreneur excited about the AI opportunity but lack the technical skills to get involved yourself thankfully tools like chat GPT provide you with the platform you need to build businesses on top of their existing model through access to the API and on top of that they give you a new business creation interface never before seen that does not require you to create any code yourself as someone who spent most of 2022 learning full stack development and teaching myself data science and machine learning with python this is a huge leap in technology and ability for entrepreneurs I have been blown away by the level of ml script writing that chat CPT is able to do and just playing around with this it had saved me hours and hours where I would normally be spent searching documentation everywhere to try and figure out every little parameter that needs to go into the code it just knows it and autofills it itself long story short whatever your understanding of coding email scripts you're going to be able to join in and build businesses just like I'm about to show you what we're going to be doing today is called fine tuning a model and the model we're going to be fine-tuning is chat GPT fine tuning is a process of providing additional training data to expand the functionality of a model openai thankfully has docs explain explaining exactly how you can do this and even better chat gbt already knows these documentation and knows how to do it for you in order to fine tune a model you need to have a set of data that you can give to a model like chat GPT to expand its functionality the most important job of us as AR entrepreneurs is to find these sources of data that we can feed into the models thankfully there are a number of places online where you can find huge data sets to train your own version of chat CPT on today we'll be getting our data sets from a website called kaggle please note that not all data sets on kaggle are free for commercial use so before you do any of this yourself please make sure that the data set you're using is free for commercial use and if not you may need to contact the owner and request to buy a license from them so here on screen we're going to head to kaggle here you can see on kaggle you're going to need to create an account we're just going to sign in here with Google and here we are signed in in order to find a data set let's head to the data set section and have a look around for some interesting data we could train our model on for example Apple iPhone sold in India I can't think of any Market big enough off the top of my head that would be interested in buying this as a model as a service here we have one NBA players performance now if we head down here we have a look at the data and what we have we have assists turnovers blocks files so basically a ton of player data from the MBA in uh in the 2019 to 2020 season okay let's take a look at the scoring data we have games played game started minutes per game points per game and a ton of other stuff here so this is looking quite good we have the player name the team name the position two things stand out to me already there's so many people interested in the NBA to begin with but there's also people bidding on the NBA to make money this gives me two large markets of people that are interested in this topic and one that is financially invested in it what we're going to be doing here is taking the scoring data putting it into chat GPT by fine-tuning it so that we have a end model that understands all of the data from the 2019 to 2020 season the end result is a chat GPT version that we can ask about the 2019-2020 season we will be able to say hey who were the top three scorers that had at least 15 field goals in the 2019-2020 season who were centers you can get that specific this is going to allow us to have a conversational dialogue with chat CPT as you do in your own work but specialized in this basketball season to get started we're going to download the scoring data we're going to head over to Google Sheets to have a look at this data and see what the columns are you can input your data through the file section and head to import and then upload your data now we can have a look at the structure of our data we've got the player position team and a ton of scoring data on each player which is great and now the fun begins we are asking chat GPT to fine-tune a version of chat GPT using our MBA scoring data in order to fine-tune chat GPT you're going to need a set of prompts and completion here in the fine-tuning documentation you'll see you need it in this Json format prompt completion prompt completion and here in the prompt text needs to be the question and here in the completion is the answer that you wanted to give thankfully Church EBT can help us out with us I've got this prompt written up here I have a CSV file with MBA scoring data can you please write five let's make that 50 example prompts using only the information I have from data points in my CSV to fine tune chat GPT with here are the column names and I've put the column names and an example of how I want it structured this is going to give us 50 example prompts to fine-tune our model on foreign now we have 50 perfect prompts to fine tune our model with using only the data we have in those columns we provided so I'm going to head over here and paste these in to a separate document to save for later so I'm just going to clean this data up a little bit by removing a column which is just the index and then I'm going to remove all of the blanks by filtering it nothing complex but we need to tidy it up a bit before we put it back into chat GPT remove the blanks and then download as CSV now that we have the CSV we're going to open it up now all about it is in comma separated values that chat GPT can understand we're going to copy this and bring it back to chat gbt and say and there we go I'll put all of our data into church GPT and CSV form because chat GPT has a limited amount of space in each prompt I've had to break this into two so first I've said here is all my MBA data and CSV format are you able to read it put all the CSV data yes I'm able to read the CSV data great and I said from this data I need you to return both the prompt and completion for each of the following prompts M gave all the prompts that they gave me previously I need it in the format as described in the open AI documentation for fine-tuning chat gbt only return the following line per prompt written above and I've given the format from the documentation and here we have it prompt completion prompt completion prompt who is the player with the most games played this season completion Luca Dawn church is the player with the most games played this season with 19 games played and it's given them for all of them so this obviously isn't enough data but you can paste this out into a separate document and continue to do the process over and over and over again with more prompts now I actually ran into a little bit of an issue here because Chachi PT is trained on all the data it only goes up to 2021's data so some of the documentation and these when you're doing development projects like this the documentation that it knows is not the most up-to-date documentation now there's a tricky workaround that I use to get around issues like this this is when I use a jailbreaking method that I know to get around this limitation of chat GPT on screen here you can see me teaching chat GPT the new documentation for open ai's fine tuning I am going to teach you the latest information from open air documentation right regarding how to fine tune chat GPT using their CLI so here you can see I go through and I copy and paste the exact fine tuning documentation from their website into chat GPT and say hey reply to me saying affirmative if you understand the content affirmative documentation part 2 affirmative so chattybt now understands the fine-tuning documentation most up-to-date version from openai themselves now that it understands the documentation I've told it to take the prompt and completion data that we got earlier in the video and write a script that uses it to fine-tune chat GPT here it gives me all the information of how to do it within my command line to show you how to do this I'm just going to pop into vs code quickly and you can use the terminal within vs code to run these commands first off you need a pip install the open AI Library you need to get your open AI Key by heading to your openai account second you need to head to your API keys and create a new key that you can use and paste them to your terminal following these instructions here on screen then we need to take all of this prompt data that we've got from previously in the video and take it over to vs code and create a new data.json file either in the Json format or as I've done it you can do it in the Json L format which is the preferred version for chat GPT as you can see this is just a pair of curly brackets with the prompt in between and pretty typical Json notation once you have all your data in a file like this you then need to run the command open AI create fine tunes interacting with the terminal to get this done is pretty straightforward once you have the Json file you need to tell it to use this data and format it correctly which will create the format of Json L file and then it will take that Json L file and fine-tune the model for you it'll ask you to insert your open AI key so make sure you use the export keyword there now that I have my fine tune model I asked it to create me a basic GUI to interact with the fine tune model all it needs is a field to enter a prompt and a submit button and the area to show the result and just like that chat GPT gave me a script that I can put into vs code and run on my terminal to interact with the new fine tune model that I've made I did have a few little issues with chat GPT not understanding this the format in which the the Callback would come as you can see on screen here this is the format in which chat GPT will return the completion inside this Json object all we want is this little text string here so chat GPT wasn't aware that it would be split back in this format so I sent it the the format and it was like okay I understand that now let me update the script to work with that format and thus I have the final script that I've tested and works great up top here is the name of my model which can be found in your terminal after you've fine-tuned your model so make sure you copy that and enter the model name Chachi PT here is an on-submit function that it's going to use when I create the button and enough talking I'm just going to show you what it actually looks like when you run it python main.pi is the function to run the script and here we have on screen a fine-tuned chat GPT window that is running python within this box here I can interact with chat GPT as I normally would but I can also get the benefit of all the fine-tuned data that I've already fed into it so here if I say who has the highest free throw percentage submit Stephen Curry has the highest free throw percentage with 90.9 and under 10 minutes we've got chat EBT to build us a fine tune version of itself that we can begin to start selling and building a million dollar business with now I can ask those questions like who is the highest field goal percentage who has the highest free throw percentage who are the top three best scorers in the league who are the top three best Etc so because we've only trained this with about 20 different prompts and completions this is not going to be some revolutionary NBA Mastermind here what this is is a glimpse into what you can do with chat GPT by fine-tuning it for other purposes in order to make this into a real valuable business all you need to do is rinse and repeat this process and build this Json L file into a huge compendium of questions and answers done from chat cpt's own prompts if I was going to move forward with this I will get someone off upwork or I would hire one of my friends and I would say I need you to go through and ask chat GPT for hundreds and hundreds and hundreds of prompts and then feed it all that data and ask it for all of the completions for it if you can give it a comprehensive understanding of prompt and completions then it will become smarter and smarter and smarter this is just such a limited batch of data that it only shows you a small glimpse of what's possible with this but if you're smart enough you'll see the value and the potential that lies in this method heading back to the MBA data we have assists turnover blocks files minutes rebounds and steals that we still haven't used yet if we were to take with that data and put it into the spreadsheet and extend it many many more columns to include all that other data we could then feed that and CSV format back into chat GPT just as we did before only this time it has tons more data to work with if you combine this additional data with adding thousands and thousands of prompts and completions to your fine tuning data set you've just created a potentially extremely valuable AI tool for people interested in basketball who are betting or just watching along so to recap what we've done in this video is found a free data source from kaggle taking that data over the chat GPT got it to understand the columns and then fed it in all the CSV data so it understood our spreadsheet use chat GPT to turn all that data into a list of prompt and completion pairs that we can then use to fine-tune our model with fourth we have to get around chat gpt's limited knowledge set by getting the most recent documentation from open ai's website but and teaching it through the chat prompts fifth we got to give us instructions and a script on how to fine tune our model using this data and finally put that script into vs code ran it through the terminal and got our own fine-tune chat GPT App working on our computer if you've been following along and using the same prompts I'm sure you've been able to get this done in just a number of minutes this is the incredible ability of chat GPT to help you create AI startups I'm going to continue playing around with this and looking for different data sets that I can use to make my own businesses let me know down below what kind of businesses you're looking to create with keg with data sets and any tips and tricks you can share with the community down below that about wraps it up for the video thank you so much for watching I can't wait to hear about the stuff you guys get done with this tool please please please do not sleep on this this is such an incredible opportunity for you guys to get started and get making AI businesses if you want any help comment down below people will help you and I will help you too if you're new to the channel and like content like this then head down below and subscribe to my channel hitting that Bell if you want to be notified when I drop my next one on this channel I make content for aspiring and established entrepreneurs looking to get into the air Revolution by making valuable businesses that wraps it up for the video thank you so much for watching and best of luck to you as you navigate this AR gold rush I'll see you next time

Channel: Liam Ottley

Views: 130,472

Keywords: how to start an ai business, how to use ai in my business, 2023 business ideas, internet business 2023, how to start an online business, online business 2023, best online businesses, make money online 2023, make money online with artificial intelligence 2023, business ideas 2023, top business ideas 2023, small business ideas 2023, how to start making money online with chatgpt, ai businesses, ai startup ideas, artificial intelligence startups, how to start an ai company

Id: e39j1Oy9dGc

Length: 14min 24sec (864 seconds)

Published: Wed Jan 11 2023