the ideas are mine but the hosts make it more presentable

@@amraswoke6762 my friend, i built a quicker version of this first before building this one for the video. my first version had full authentication, bcrypted passwords, jwt token verification for the api... but as I say repeatedly in this video, the features are incomplete in this app because the point is not to build a full set of working features. in this demo the authentication is not set up and the passwords are not encrypted because the app doesn't matter and the features don't matter. the point is that i built a clean full stack app architecture ready to iterate on and build out, while writing 0 lines of code myself

@@amraswoke6762 many other features are unimplemented in this demo, as it's presenting an overall building workflow, not presenting how mind-blowing a horse racing app is

@@0xb1sh0p8 the video demonstrates ai replacing upwards of 90 percent of my traditional job as a full stack developer, in 2024. you are interpreting "replaces full stack developers" with "replaces ALL full stack developers by the end of 2024" which is not what the title says

@@shawnfromportland lol..Ok..now build something with some actual complexity and record it without edits. Gonna be a long ass video and burn a lot of credits.

@0xb1sh0p8 most of the code was generated by codegen software not llm, with accuracy and token credits in mind

@@shawnfromportland Lol your title is literally "AI Replaces Full Stack Developers In 2024" you're being so disingenuous. You did NOT title the video "AI will change the way Full Stack Developers work in 2024". 🤦‍♂

@@TheDMHP would you a like a refund on your channel subscription?

everything is 100 percent AI-generated. i spoke a casual rambling discussion of my thoughts on the future, and gave the transcript to NotebookLM to generate the podcast.

i demonstrated here ai doing 10x as much good development work as compared myself unassisted. this serves as proof to me that developers in the traditional sense are outmoded, and while we will still have jobs they aren't going to resemble what we were used to before.

​@@shawnfromportland You are comparing to yourself, retard. Market don't care your opinion.

@@aaronvadnai almost there

@@vasilijemilenkovic1069 you must have missed at 7:39, i said that if we can assemble something like chatgpt o1 + Wolfram, that's pretty much AGI by my definition. other people have their own AGI definitions 👍

@@13odman in the demo app i was building which i will present in the next video, i used Claude, ChatGPT, and copilot

I found building one makes me think of more things I need as I code the thing to help me code 😅

i agree it's not "all" AI causing decline of tech jobs, but i suspect it's more of a factor than most people are acknowledging so far, even if it's just AI FUD

I'm open to it. what about specifically? "it's bad" is my take 😂

@@shawnfromportland fair point 😂 I guess I’m wondering if this is the worst you’ve ever seen it or if it’s been worse than this through the years. Essentially where does this one rank lol

@@sillybilly346 it's the worst tech job market i've seen. i've applied to 225 jobs in the last 6 months, had 5 or 6 interviews that sometimes went to a second or third round, and no offers. mostly ghosted. this time also feels worse than past tech job slowdowns because of the outlook with AI in the near future. These are unprecedented times. 2008 financial collapse felt similar for a while in the jobs market, but in a year we seemed to bounce right back. just 2 years ago, anyone could go through a few weeks of code bootcamp and learn basic js or python and land an entry level jr. web dev job for 60k+ pretty easily, and today it feels like those days are gone perhaps forever

Where is the code of the complex app? This is like a lie unless you show the GitHub repo.

these videos i will be posting are in the spirit of thinking about what's coming moreso than what exists TODAY. i dont know what management looks like in the AI age and how that fits into the team, we dont even have an accepted standardized AI dev workflow yet. Right now i'm just trying to get into the mindset of the developer of the future. by "end users", do you mean web devs or website users?

​@@shawnfromportlandgeneral public end users. It is jaw dropping accuracy for us because we know what it is doing. And corporate management is a powerful class of hands off jobs.

My soapbox atm anyway, is that code is a job for computers. We may even start seeing futility in high level developer friendly languages.

@eatfrenchtoast i totally agree, code is for computers

What would be fascinating would be you explaining more about how to practically setup that "the world revolves around Swagger" ecosystem you mention at the end.

@@ttaylor9916 that's the next video, i got pretty far with a prototype of this workflow before filming this video, but i want to get it together more before presenting the code. but come back to see it

thanks for watching and commenting. **Are we imagining that a regular person, who doesn’t have much computer science knowledge, could use AI to build a full-stack website? ** as it stands today? no. however i’ve seen there is no technical limitation to making that a reality today, it’s just that no one has built that tool yet by assembling all the necessary technologies into a nice package. for example if we tuned an AI to prompt the nontechnical user to ask the right questions, then it could be achievable now to have a nontechnical person build a full stack website after answering a hundred or more questions. **for a layperson, if bugs show up, they’d probably just keep asking the AI to fix things, which could even cause new issues.** For the AI we have today, yes, thats exactly what happens, and is a common issue ive come across in my experience generating apps with ai the last 2 years. I’ve been continually trying to find that dividing line with tasks between what is less work: to go thru the ai, versus just doing it myself because of all the AI bugs i’d have to deal with. It is a moving goalpost, and by the time i finish one project where the ai was not capable, there are loads of new AI technologies or techniques available to try on the next project where maybe it is capable the next time. **Even if the AI eventually makes a “working” site, there could be unused variables, extra functions, or other things that aren’t ideal-but the person might not notice these problems as long as the site appears to function. ** I believe that a) that’s a mostly anthropocentric viewpoint, coming from the world where it is humans who are primarily looking through the code, a world in which we are embarrassed during code reviews if we left a bunch of messy comments and unused code in our app by accident. no machine will care in the slightest if there are unused functions or variables resulting in a couple kilobytes of extra space and the desired result is still achieved. eventually, we won’t care either except in only a handful of cases of highly sensitive code (life support machines, nuclear reactor code, etc). and b) this is probably an irrelevant point in the future when ai is sufficiently advanced. if it’s good enough to do what it already does today, then it’s good enough to identify cruft and unused code as like a discriminator layer. it makes the unused code and messy code errors largely because it’s a reflection of the person commanding it, today. try taking some of your more messy code, pasting it into claude or chatGPT, and asking "please point out inefficiencies or unused lines in my code." it will happily do it for you **building a reliable website requires some understanding of system architecture-knowing how databases, servers, and APIs work together. Without this background, it would be challenging for a layperson to build a secure, robust system, even with AI’s help.** again, that applies to the tools available today, not to the tools of tomorrow. notice that technically, if the layperson was motivated enough, they could prompt the AI to tell them what they dont know and give summaries of their blindspots. in fact this is my starting point today whenever i jump into a brand new deep technology i know nothing about and i quickly want an overview before studying more closely myself. **Getting useful results from AI often depends on knowing the right questions to ask or having a basic idea of what needs to be done. That’s something most people might not have yet. ** a layperson could get far just with what exists today, by prompting “im a nontechnical person that knows nothing about building websites. im interested in having AI help me build a website. prepare a 1 page report briefly summarizing the main domains of knowledge i should get more familiar with, as well as an overview of AI prompting strategies to help me achieve this”. but that’s just what exists today. I’m trying to think a lot about what’s coming rather than just what we have today. i built a prototype app following this outlined approach of this video before shooting the video, but im totally aware its only a matter of time (and probably not long) until an AI model is also better as the orchestrator of AI than I am. thus even the ai-maxxing strategies i’m talking about in this video are really just stopgap measures until AGI and beyond. **Just in web development alone, there are a lot of frameworks and architectural decisions to make, and while AI might eventually handle much of this complexity, I don’t think we’re there yet.** one use case that Claude works incredibly for, is pasting in technical documentation that it hasn’t been trained on, like a brand new framework, and then asking specific questions in reference to the docs. It is already at expert human level in this task. and even with having no knowledge of architecture, you can ask the AI for architectural input and commentary at every step of the way and you are already going to generally get very good advice. sure, chat bots of today dont build a full stack webapp on command in a prompt or two. but think of AIs which are finely tuned orchestrators, commanding unlimited other AI agents specializing in any other given framework or topic. crude versions of this have been built already (autoGPT) and the results are pretty astonishing. it’s a matter of time until this is commonplace. **This reliance on prediction over understanding is a major hurdle in achieving true AGI, as true general intelligence would require a model to see beyond data-driven predictions and exhibit genuine insight, adaptation, and reasoning.** try playing with the o1 models. some level of reasoning has now been added into the models. its not perfect, but its very impressive. also note that prompting strategies alone can yield the same results as the cutting edge reasoning models on the old models, simply by prompting “first come up with a plan and detail your steps to accomplish this before doing what i ask” (chain of thought prompting). again, today is the worst these models will be at reasoning, and today they are damn good. **Without [interpretability], it’s hard to rely on AI alone, which makes AGI feel even further off.** ai is not superintelligent yet, it’s true. yet, even as a simple next token predictor, when trained upon global scale knowledge, its already an extremely powerful tool. getting the list of sources you mentioned and then doing the google search yourself on each of them to find out of each one is real or fake is already a valuable timesaver over just crawling the whole internet yourself on a topic. I’ve used chatGPT in this way to get leads for obscure research topics im interested in and while sometimes ive gotten fake sources, i’ve also gotten real sources from it that were very helpful. **To truly reach AGI, we’d need AI to not only perform tasks accurately but also to adapt and learn across a wide range of areas, make decisions based on feedback, and reason independently. ** at the frontiers of research we have all of that today. it is not packaged today in a convenient end-user package. **Current AI is powerful, but it’s still very task-specific, and the broader intelligence needed to handle complex, real-world situations on its own doesn’t seem within reach just yet.** keep in mind that the ai tech we have today was unthinkable before 2017

I strongly recommend the book for you Superintelligence by Nick Bostrom.

@@stingfiretube why waste time looking at noise at all, when you can evolutionarily seek something familiar like stable diffusion

@@MaxiToysOfficial i don't know what that means but ok!

@@shawnfromportland I was joking

​@@shawnfromportlandit's a very old but famous meme

I could probably do that but why would I?

@@piergiorgiocestra3833 ye

​@@labamba3011 it was supposed to help you zone in on the intellectual imagination journey 🤙🤙

thanks brother

@@TheParkBadBoy thanks 🙏 come back for a few more parts to the story

i mean i was making 60k which was supposed to be "ok" for the time, but after living in a big city and paying rent and college loans and eating out sometimes, it was gone, i wasn't saving anything towards a house

@@shawnfromportland Yea I was making over 70 to 80k with bonuses as a network engineer, and left. It's just soul sucking. IMO if you make enough to be ok but can have more time that is better then making more, and having no time. Time is the important commodity not money IMO.

@@coreylewis171 thanks brother, miss you 🙏

@@coreylewis171 it was a lot of fun and you added a lot to the video

@@MattCookOregon i made the skate game documentary sound like the lamest thing ever in this, i didn't mean to. id still actually like to do that one day but i sold all the consoles and games and it doesn't rank anywhere on my priorities yet. it will prob never happen. i was going to focus on the style aspect, how well each game lets you do whatever tricks you want. try to get style tricks in every game even the ancient ones

@@Roh8n thanks m8 🥂

i namedropped you in the '08 part in this series 🤙

@TheParkBadBoy thanks! i get better in the coming part reviews

i was kind of talking about myself. at 41, if i worked for 12 hours and skated and played shows, my body would break down and i would be exhausted and probably couldn't keep it up for more than a day or two. This is a fact of aging, it's not a judgement of the next generation. even Kobe Bryant had to get out of the full time game by 41

thanks for watching

thanks mayne. I've moved to upstate New York because i wanted to buy a house and that's impossible out west. check the rest of the series! more parts to come

@@lovemycamry thanks for watching and sharing that.

@@jamesplacencia6741 thanks mayne

@@staylame ♥️♥️