Claude 3 Opus vs ChaptGPT Plus (mind blowing!)

hey folks it's St From Evolution Unleashed coming to you today with the cut lip but the show must go on today we're going to dive into this big announcement from anthropic or the release of Claude 3 their new model and they are claiming that it is uh the most powerful model on the market the new Opus model apparently outperforms gp4 which runs cat GPT plus across every metric now I'm a we bit cynical don't know about you but Google recently launched Gemini Ultra and said it outperforms across benchmarks as well and what do you know it was an unmitigated disaster their uh text to image generation wouldn't make white people went overly woke they're probably going to fire their entire diversity team and it was absolute rubbish and what do you know it did not beat chat GPT at all I did a video on that uh probably a few weeks ago which if you scroll around you'll be able to find or I'll link it down in the description but today we're going to put anthropic to the test are they being like Google and just fine tuning on the benchmarks but really don't have enough to back it up or have they actually developed a more powerful model let's go find out so what we've got for you today is a comparison document and this is yours free I've dropped it down in the description of comments depending on where you're seeing this video and it is going to show you I give you all the comparisons the prompts I use the outputs and whatnot now I'm not going to read this out to you today folks what I'd rather do while you're watching this is give you my thoughts and insights into the differences between the two with the ultimate goal to figure out well which model is better for what tasks and if you could only pick one which one would it be and why and I think that is more useful because we're trying to use these tools to create impact to move us forward so let's go take a look so there's a table of contents here to quickly navigate and the three areas we looking looking at as with Gemini uh we're looking at article writing target market research and education and learning so you can just click here you can have a look at the prompt and the outputs and what we did is we took the first output from the prompts in the raw form pasted it here we tided up the headings but we didn't edit any text at all so we begin with our article writing prompt now I won't read this to you but this is uh a professional prompt that has been engineered for copywriters to create a lot of articles we've got copywriters creating 500 articles a week with this because it creates amazing articles the power of professional prompt engineering folks and if you want this prompt now this one I've already filled in all the placeholders for the purpose of this but over here slight tangent before we dive in here's another free gift for you the new Ultimate prompt toolkit I finally got it done after four months of working on it it's a couple hundred pages long and this has all our professionally engineered prompts available to you for free and it includes our right this one here writing epic articles using chat GPT that has all the placeholders so that you can use this if you're a copywriter and there's a ton of other really cool prompts here as well folks um I won't go into that but the ultimate prompt toolkit is yours there's a link in the bottom of this document so you'll be able to get your hands on that so let's go look at the outputs interestingly uh the word count Opus from CLA 3 came out much more words 764 and one of the problems that AI models have had up to today is that they won't write the length that you want so weirdly this time uh Opus gave exactly what we wanted 1,000 tokens 764 words and it went bang on so hey finally we might have a a model that can give you a word count I haven't done enough testing to know for sure but uh it's a good positive start for plaude okay so what I liked most between these two in terms of the actual article what I got it to write about was the power of personalized AI it's a topic dead in my heart right now folks and I really like the title uh much better from Opus or Claude personalized AI your kid unlocking Limitless potential and it went through and it it sort of gave it all up so where did each model shine well actually I have to say entirely that Claude 3 uh just wiped the floor with gp4 here folks and for me to say that is the first time I've ever said there's a model better than gp4 when it comes to writing so what I liked about it was it was more nuanced it gave a lot more sort of Rich detail it really explain Concepts uh at a what I consider a deeper level which previously was a strength of gp4 and the gp4 article is still good the problem with gp4 is the way it writes we all know and love the output from from GPT don't we or from chat GPT it's very robotic it's got a very um easy to spot writing style and sure with prompt engineering you can mitigate that somewhat but you can still recognize that it's chat GPT if you're not editing the output of course folks you should edit all AI output before you deploy it to the wild because these models hallucinate make mistakes you should be editing every piece of content anyway you're not here for a lecture are you that's for our prompt engineering course tell you all about that so went through you can have a read these but I have to say really impressed with pla 3 nuanced details uh really tied it together nicely but more importantly it spoke to me as the reader and it really stayed true to the topic and I really uh enjoyed reading it so hands down CL three beat gp4 here folks okay so that's writing and we always known that gp4 is a bit weaker in writing style so what about research where gp4 was miles ahead and again we gave it The Identical prompt uh basically we gave it two pieces of information um my high level Market is business owners entrepreneurs and high paid Executives that are too busy and my offer is Fitness coaching where I help them overcome their last barrier of success their personal health in a way that is easy to build into their busy lives and so on you can read these prompts and again that prompt guide that I just told you about our toolkit has all these for free as well so you can use them in your own business or just to play around and see which model's better okay so we got Alex Priya and Michael three personas we're GNA focus on Alex oops that was a mistake oh what have I done let's go back gosh we pressed the wrong button but I'm not re-editing this video so you'll just have to get it whats and all or cuts and all today anyway I D let's look at the values and beliefs I really like how chat GPT wrote this you know uh belief in the power of discipline and hard work gave really good crunchy useful information and as a marketer myself I can understand the value in this information I thought it did a very good job it wiped the floor with Gemini and uh I think it's going to be pretty hard to beat when it comes to research so let's have a look at what uh Claude had to say about that and my goodness okay John Smith Claude come on be a little more creative with your names Emily Johnson Michael Lee the most common names you could find but they're just avatar names we won't penalize it too hard for its naming you've seen the way I name my products uh yeah I'm I'm definitely not someone to judge but when I had to look at the emotional drivers the values beliefs the objections which are the core crunchy stuff you need in marketing to align your messaging effectively wow much more nuanced right bang on uh again Bo three excelled here folks and I mean exceled it talked about the emotions it talked about the drive what really made them tick at what I thought from a marketing perspective at least was a deeper level but I also had it do one other thing and here's the GPT one I asked it to create an aligned marketing message or a unique value proposition for each of those personas so a uvp is just basically how we communicate our offer in an outcome focused way to our market so they understand what we do and more importantly what's in it for them talked about that yesterday's video okay and I love this I'm going to read one out here because it's so good transform your health and the time it takes to listen to your favorite podcast remember Alex is into podcasts wow bang on absolutely bangging marketing uh our bespoke Health Solutions speaking to that higher sort of level executive bespoke uh fit seamlessly into your entrepreneurial lifestyle empowering you to achieve big performance with out compromise love it I think that is powerful effective messaging I tweak it a little just to fix the language but very very good unique value proposition so let's have a look at John who's into golf reading business books and travel and let's have a look at what uh the marketing message alignment was like because that alignment is a critical skill in Ai and I value it heavily right so this is an important indicator gain an edge in your competition and Achieve Peak Performance by optimizing your health are already I'm turning off so very interesting here and when you look at all three uh gp4 much better at creating a unique value proposition and aligning that to the market and that requires a far more complex sort of amount of programming or abilities reasoning whatever you want to say so I think gp4 still has an edge there and for me that's pretty important so I won't be giving up gp4 anytime soon but if that's a sacrifice if that's the only sacrifice I have to make that would be one I'd happily make if I could only choose one of the other because the research was so good from Claude so this one's a bit tougher Claude had better research but gp4 had better alignment to the market and the one that's going to make you the most money is the alignment to your market and the gp4 market breakdown was still very very good now imagine if we took claude's output and got gp4 to create align messaging we're probably going to get the absolute best we can possibly get out of AI doing it that way anyway last metric I don't want to keep this video going too long education and learning what I've done is I've given it my my brick score very simple and said teach me time management in a way that an inpa would understand if you want to go find out your own uh personality score folks Here's the the the link here 16personalities.com entirely free you can get your mbgs indicator and uh very very cool uh and then you can can actually align AI output to it obviously we go much deeper with project you decoded but this is a good introduction to that concept so basically we're asking it how to incorporate time management into busy routines and explain it to a beginner in a way that I can understand and again this time C GPT gave much more output 566 words versus Opus only gave 391 but when I read them both now this is where I can be entirely subjective because I asked it to align it to my writing style okay and we just talked about the alignment from the marketing perspective we thought gp4 did better but in terms of explaining information rather than marketing which is a different skill set but a similar sort of functionality I guess I really was impressed with Claude but it also threw up a couple of errors or a couple of issues that I think deserve a bit more sort of diving into before we declare it a winner I really liked that it understood how to link my personality style to time management I think the thing you can read these yourself but I want to just pull out one thing that it said really that that blew me away uh where was it must be right at the top so it basically said yeah uh to where was it yeah here it is a to-do list will help you complete important items while calendar alone may allow you to meander without accomplishing priorities so it's just saying use a basic to-do list that resonates with me very strongly I thought the information was very much uh the sort of information that I would want but here's where it went a bit bit off the rails um so here's the exercises to reinforce time management now I know it's probably just needs a bit more I would need a followup R please expand on these to explain it would probably do a good job but part of this test is to give the first outputs and see how good they were and its exercises to reinforce time management lessons was a bit odd practice prioritizing tasks by filling a jar with rocks pibes and sand in the optimal order now I know it's just speaking about an analogy it hasn't sort of broken it down and explain that that it is an analogy so if I was to follow this literally I'd be out there getting a jar filling it with rocks Sands and Pebbles and it wouldn't mean anything and the same thing here you know spend $86,400 as your seconds in a day but not really giving me anything tangible whereas I felt gp4 was a bit more tangible though I didn't think it understood me as well what I mean by that is it did understand me but it was trying to create lesson plans and exercises and stuff like that but it did you know understand a lot of the stuff about me so this one I think was Clos and I think I'd be happy with either output if I'm completely honest uh but I thought that the way C wrote it was a lot bitter and it really did resonate until I got down to filling rocks with jars and pibbles the last thing I did is I hastily wrote a conclusion in here and uh I probably left out a couple of the points I've met uh made today but I just asked each model to rewrite the conclusion identical word counts pretty much two words difference and Claude uh wrote better again chat got a little bit too uh here's the de uh and that's totally cool we want to find the best AI buddy that helps it knock it out of the park now I'm not sure why it's gone to that mode but it did whereas we didn't get any of that kind of try hard bro stuff from Bo anyway that aside uh you can read the conclusion for yourself folks uh and then before I sort of wrap up here's some additional resources for you we've obviously got the uh the free uh ultimate toolkit prompt tool kit we've got a free custom building guide custom GPT building guide if you want to build custom gpts on chat GPT free Pro you decoded going into your uh wiring and DNA to hugely powerful stuff to align AI to you and who you are massive breakthroughs possible how to make cool Studios like this and so on and so on so you can have a look at that folks the the link is below but I'm really Keen to hear your thoughts have you used Claude 3 uh where have you found compared to what you using it did well where did it dro the ball let's sort of collaborate share our sort of feedback and experience using this model so that we can understand you know exactly how to use each of these models to the best of their capabilities so that we can create real world impact remember we're not here to create just jokes and silly stuff um you know that that shallow content we're here to create real world impact use AI to give us the biggest Edge we can over our competitors especially moving into this transformational era so for me at least understanding which model is going to maximize my Edge is very important so be very keen to hear your thoughts what you found you like or dislike about both of these models let us know in the comments and uh that's it for today so hope you enjoyed that little comparison I'll see you tomorrow in the next video peace

Channel: Evolution Unleashed

Views: 535

Keywords:

Id: niZyONSYe4s

Length: 16min 18sec (978 seconds)

Published: Wed Mar 06 2024