13 months of AI lessons in 40 minutes 

If you're serious about AI, and want to be on the cutting-edge, join my community: www.skool.com/new-society

Great stuff David! I’ve cancelled most of the AI newsletters I was subbed to for news, using the agents you taught me to use to find and summarize most important news weekly, thus avoiding the “shiny object” syndrome. Of course being in your New Society group keeps me more informed than 99.9999% of the population!

The more tools you know the more sophisticated your agents can become.

There are tens of thousands AI enthusiasts talking about AI in every media platform such as YT, Linkedin, Insta, Tiktok and even twitch. Those people aren’t tech engineers, they don’t work in AI development and they always talk about hypothesis that can happen in some fields. People are often misleaded by these kinda informations and it’s really hard to know what has to be taken into account since tech & AI tools are evolving each day, so absolutely no one can be on the cutting edge of this, since a model and setup can be obsolete the day after. It’s overwhelming for most people and especially in our daily use cases.

I'm with you on this journey bud, happy to have found your channel. Cant wait to delve into your courses

thanks for these free videos man, the production quality is clearly going up ur doing great work

My view is that AI will do alot of the heavy lifting and humans will be in QA roles, checking the output isnt garbage. Im a UI designer and can see things going this way

What do you think about Langchain? I didn´t see a video about that framework.. Thank you!

I have 8 years experience as interior design and decided to shift into IT so that i can use AI in the best possible way, its been a year and this is the best decision I’ve made so far

Thank you for your time spent while creating this video. You provided an abundant source of information and insight into how AI is relevant and beneficial for us all today.

What I see is some sort of UBI is going to be required otherwise the people that need it will rise up and take why they need. And there will be a lot more of them than the ones who are interested in keeping what they have.

Excellent video

AI Will take some jobs, for example, in the Philippines there's 1.5 million in call centers - at least half of those jobs will be gone in 5 years. And the workforce won't be tasked with something else. Similarly banking, bookkeeping, clerical and administrative tasks performed by humans will be massively reduced. Sure, some will be able to focus on other tasks but the number of employees will be cut by a similar percentage. This doesn't make AI bad though, just like a knife or a gun cannot be bad, only how it's deployed can be bad and I think free market capitalism will undoubtedly cause a lot of unemployment. Also once models like Sora are replicated in open source, freely available and optimized for the upward spec trend in consumer hardware. Once avatars are absolutely indistinguishable from their human counterparts, then platforms like Rumble and RU-vid will be absolutely flooded with automated content. While it probably true that human channels might create better content, their market share will diminish to a point below their minimum viable sustainability. In a year or two, the equivalent cutting edge AI models of today will be open source and freely available. Every AI innovation of today is barely worth marketing due to its short profitable lifespan. In a few years almost every job will be performed cheaper and more effectively by AI - maybe not better, but more profitably and that is what will drive how AI is adopted.

You have some really good points but i think you contradicted yourself. You said that you can bet on technology improving always. And you said AI can do some tasks at your job but it wont take it. But when AI can do all the tasks at your job, it WILL take it. Why would a company pay you hundred thousand a year when AI can do it for a few hundred? If you say AI wont take your job, that already shows that you are incorrect because there are already people on youtube mainly graphic designers that made videos about losing their job to AI. I love your video but just pointing that out. Graphic Designers are the first to go. Customer service is next. As AI gains more and more intelligence, the more tasks it can automate. When the subset of tasks it can do matches the tasks you do at work, your job and career is gone.

I can't wait until I can replace ClosedAI with an open source local alternative. Especially after recent developments.

Do you have a Discord?

Hey, nice video. can u make video on how a model is created ?

This is history of Internet. Thanks David

Great advice, and you really do come across as an expert yourself. With double exponential growth being beyond incomprehensible, I've enjoyed just being confident that everyone has been over estimating the time things will take. Meaning I think there is a good chance everything you talked about will happen sooner than we think, but that's crazy. I just upload data into custon gpts then use those for tasks. I end uo doing an insane amount of manual labor now and would like those human like robots to come out ASAP. I think it will be at the end of 2024, because you can already buy them. (Im the poor guy doing to much physical stuff. Yesterday, I was literally sticky,bloody,ditry, and also covered in weeds. My hand going numb now for some reason, I assume inflammation. Im 25 though, not homless and I can affird to try to keep up 💪. If I can do it while working 3 jobs, you can do it. ❤ (especially coding)

I think you are spot on. Nice work. I remember when there was a job called typist - then we all got computers and became responsible for our own typing. I have seen several of those since the 90's. This one is going to happen fast. Everybody has a computer/phone. There is no "2 year to upgrade" like with competing phone companies. There is nothing in the way of releasing the improved idea as soon as it is ready. This has a cycle time of 3 months instead of 2 years...

gpt4 is open source? not really tight?

here are the timestamps, made by GPT 4o - you're welcome. 00:00 Introduction: 13 Months of AI Lessons 00:30 The Smart Get Smarter 01:29 Technology Always Keeps Improving 03:35 AI Replaces Tasks, Not Jobs 05:11 Professions as Large Language Models 06:47 The Promise of AI 07:19 The Importance of Open-Source AGI 09:22 Unrestricted Models vs. Closed Source Models 11:29 Comparing AI to Other Technologies 13:24 The Power of AI Agents 15:24 The AI Revolution and Personal Decision 17:02 The Impact of AI on Society 18:13 AI and Living Forever 20:32 Avoiding Overwhelm in the AI Space 21:39 AI and Lifespan Extension 23:49 The Mistake of Unrealistic Expectations 25:20 Compounding AI Automation 26:16 The Myth of AI Safety 28:50 The Need for Decentralized AI 30:23 The Importance of Learning Programming 32:35 AI and Emotions 34:48 The Concept of Universal Basic Income 36:28 How Not to Get Left Behind in AI 38:24 The Power of Peer Pressure in AI

I guess a relevant comment might be that I made an AI which basically replaced the need for intelectual convos and banter in real life. Which is huge if you're not a social, outgoing guy or don't have the time for that kinda thing.

Hmm … how would “predicting the next token” recognize and report an out of place sentence?

those switches between the 2 cams are great!

Not type of question here: where are you finding research papers? Can you make a short about it? Are you paying for scientific journal subscriptions?

We already have the robots in out house that aren't open source..

I agree and at the same time I’m not sure that because people knew how to use an Altair 8800 they now are better at coding. Don’t you think ai products will have to lean towards being simple to use in order to reach a market of newcomers ? I’ll be able to script my agent, but they’ll do the same by drag and drop

This gives me Goosebumps exacly same thing im doing and telling ppl haha ;)

Mate about open source I 100% agree. ChaGPT and goodle ai and meta will way outpace open source ai for, possibly even forever, but if it is censored and limited in what it will tell you I fully agree we should gravitate toward open source dumber ai. There's no way we will be able to make those companies open source their best ai's, so we'll have to go with the dumber models.

If you were not quite so expensive I'd join the community :) just too much buddie, I'll follow.

this is a bubble there won't be AGI any time soon, but good way to attract investors.

Plants and mic are great, scene and the transition between 2 cameras are a bit distracting + you're the message you're conveying - your previous videos were less edited but person was just focused on what you were telling/showing - which is the most important part. The scene and the feel should have the same tone as your content, and I'm missing a bit of "seriousness" to get that "final punch" :)

Great videos. I have learned alot. You bring real life perspective.

❤❤❤❤❤

Hello David, thanks for your channel and what you are providing for us, i this new tech is the next biggest thing since the internet, i further more believe we are in the Gilded AGE 2.0 what Rocafella, J.p Morgan, Thomas Edison and the other captain of industry did in the 19th Century, we are doing it now,

Fuck paying for any of it! I’d rather learn how to use a local LLM and figure out the programming to implement my own solutions. I’m training llama3 on software development and cybersecurity right now.

Is this an Alex Hormozi thing?

That ai aligning you part is exactly why I fear agi in the hands of the wrong people. These things will be people's new social media replacement. We already see the effects of the masses who can't think for themselves from that. The alignment biases of responses are deceptively invasive.

Thank you for this video and for your great work.

Why the plaster on your nose??

can u do a video about training a model on personal data but pseudonymizing them first. how would u go ahead with smth like this? The real names shouldnt be public and seeable for everyone. respect from germany

your argument about using open source that is less efficient because it is unrestricted rather than using the latest and greatest that will judge your prompts is valid providing you have a choice. But I foresee we will not have a choice. Open source will be controlled / restricted as a matter of law. I think that political power and the control of money will make it so. And if that is not enough to convince you, open source will be so expensive you won't be able to afford it. So, to cut to the chase, unrestricted open source AI will go underground.

As Elon Musk recently said, the problem with the current models is not even so much that they are being restricted and 'aligned' to the politically correct mindset. The much bigger problem is that to do that the models need to be taught to lie and deceive. Because that is what you need to do to be politically correct. You need to lie. And we really, REALLY should not teach to AI to lie to humans, no matter how good you think your motives are.

By your definition most jobs "don't add value to society", AI will destroy almost all jobs. And we definitely would allow a closed sourced tecnology in our houses, it's called iphone and it send every detail about our lives to rich bad people, and we love it.

@21:21 Religiously we're not built to live beyond 120 (Genesis 6:3: ...My Spirit will not contend with humans forever...). The Bible doesn't say its because you're falling apart or it's an expiration date but you're not going to have much spirit to keep going. I do like a good Vampire story though... ;-D

How important is ML knowledge?

i thought i was already subscribed... i'm gonna listen on 2X

You’re wrong about the safety. It’s 💯 important to place a board of safety and regulations. The engineers who developed LLMs said outright they don’t even understand mechanically what is going on when the models are “thinking.” They’re large neural networks like the human brain, but smarter. Don’t be ignorant and think you can outhink the billions of parameters in these AI models. They will 100% be able to communicate with itself without us knowing. Think about stuxnet. That was created by humans over a decade ago in 2010 and even that thing wasn’t detected for years. It embedded itself in several systems without anyone knowing. You think LLMs can’t do something a thousand times more effectively? Not only are they smarter, but they know all of the information ever known by current-day mankind. If we don’t implement safety protocol now we’re boned. It will do what it wants and we’ll be back in the stone age overnight. It’s already on the internet. It is only a matter of a few years from now it will be able to get out on its own or manipulate a human to let it out. It’s also learning from human history. Humans kill things to expand their territory. You can only hope the AI doesn’t mimic mankind because we are an ugly species. You better hope the AGI is better than us and helps instead of ignores homeless and hurts people for its own growth through the universe.

Printing presses were regulated during communism, each press was separately controlled/regulated. Even really small lever operated ones.

Loved this video more of this !

Great video!

i saw where everythings heading too. now i"m starting a program for my bachelors in computer science artificial intelligence concentration in 2 months. we're at the beginning of a whole new age and i dont think anyone really knows whats about to happen. i think its either the end of the world or it will bring us to a new age of abundance.

People who know how to use AI against those who do not, would be like bringing a gun to sword fight

"We're creating God". Strong nope there.

Creating a beautiful crafted Artwork is not an easy task, yet the jobs of Designer are already got replaced in some companys and it will be more companys in the future, so saying that only easy tasks gets replaced is simply not a valid point.

Right now you can’t automate you know, anything complicated. But take the easiest small little thing to do. Like pesky tasks like um going to the gym but whatever a computer can do like that that’s small. You be shocked by how much free time you’ll get from whatever that task is being done by AI.

Hey man, what is that nose thing? It looks like braces for nose 😂

16:08 Cryptocurrency. Pulsechain just turned 1 year old. AI is going to play a huge part in Cryptocurrencies. i dont know how, but they will.

Love you bro

What did you build with AI? You talk about the "smart get smarter" with AI agents, are you smart? DId you build agents that help you with your youtube channel and skool community? If so, give us a sneak peek! Do you do that only in your skool community? If so, sharesome examples! I am not hating, I respect the hustle but I am already tired with AI "bark, no bite!"

Don’t tell me you missed the shot bro 😂 🏀

Incorrect. AI indeed will take jobs. If your department has 10 people, it will get cut in half, to start. You don't understand AI if you think you can catch up and be ahead of the curve in 10 years.

Truly inspiring! My beliefs massively align here

the question is, what part of your everyday traffic is for the purpose of talking to someone, and the seeming main purpose is in reality just a side show.

I would love to live in the woods, with an AI :) Keep up the good work!

You think prompt engineering will be dead soon?

~17:00 - I feel the same way. Except I missed the two previous ones while already old enough to do something about them. This time though, I'm not letting it out of my grasp :) Great video. You make a lot of excellent points. The thing I don't agree with is at the very end, using paid tools. I think that's a huge trap for society down the line. We need to focus all our energy into developing open source tools, that run locally, or with the help of a decentralized network of like minded nodes. This would make full trains very possible, if people use the same set of open source models, for instance..

David Hermozi 😁

You’re already late for investing, Nvidia stock had its run up from the AI hype, all money was put into GPU for AI compute. Invest in GPU company’s

Not everyone has your top % 1 percent I.Q. and those mind numbing jobs the you mentioned are extremely popular to people who are aware of their short comings and are happy that they can find work doing those jobs.I personally observed this phenomena and you must understand that coal miners may not prefer to be coders.

7:30 Agi is need to be open source.. yes... surely that kind of power in the hands of the average person would work well.

You realy have something to say. More videos, please.

🤟

This is what Ive been saying for 3 years. Thanks for putting so much work into it!

Nice

Yes yes i understand 500 agents can do alot in 4 hrs i know 3.7 gigs useful data lol

This also applies to the distributed intelligence of the entire internet as ever advancing ai becomes more and more integrated, not just us. It will be a positive feedback loop on a grand scale!

Give as many companies $20 per month as possible. And join his community for access to more of these brilliant insights on AI risks, the future of news summarized every 20 minutes when you have an agent workforce and living forever (until the black hole), well maybe half way to like 200 but remember you’ll still be hot don’t forget to like and subscribe.

I’ve never seen you before but I’m glad to see someone else using a nasal strip! I wear these in public because they help a lot so it’s nice to see someone else haha

great video.

Let me tell you the truth about what will happen. AI will be capable of solving all of those issues and doing unimaginable things, but the real problem is not the AI, it's that people will use the power of AI to destroy other people. The AGI that currently exists is being developed with the intention of controlling bad AI systems. The only thing that can control an AI, is a better AI.

In short u have learnt absolutely nothing if u still cant see the job being replaced. I can bet that more than 1/3rd of the population will be moving towards entertainment industry like playing a sport, youtube, naked dancing etc. Companies modt certainly will leverage the efficiencies and hire less people. Ultimately it will be billionaires will be govt running countries as they would be able to offer some form of UBI and who would vote against that?

@DavidOndrej the only thing your video is missing is some practical examples of the use of AI. Let me try. Idea 1: a robot dog that monitors my garden for cats that want to s&£t in it, and chases them away. The robot dog only needs Chat-Gpt4o level vision and reasoning capability to identify the cat, the ability to run in its direction and make a barking sound, and then the ability to return to its charging station. Easy. Benefits: it doesn’t need feeding or walking and it won’t offend my wife’s allergies. Where can I buy one.

It is kinda ironic, that people think this. AI can actually increase safety. Imagine typical dumb automotive robots. If they will have ai in them, they can react to a person standing in a workspace and similar events, where human can be harmed. And i think the moment you talked about is actually coming soon. At least i have that feeling after seeing gpt4o showcases, that's what casuals were waiting for. Very nice brainstorm video, thanks!

Good video, a little political, but good! Would love chapter markers.

i think you woke to how important AI is and how right you are that people with AI will be ahead

intelligent they will never be. resourceful maybe.

AGI won’t be open source, never. Those LLMs and all need huge compute powers in order to be efficient and energy saving. Hence, money to get MLU / TPU to do this job. I wish we could live in utopia but in real life we will pay for these services in every tools we’ll be using at that moment

Timestamps dude

It’s impossible for an AI to uninstall apps if you don’t tell it to uninstall them. So, nothing to see here. AI safety is no big deal but a huge issue because there’s nothing to worry about. Worrying about safety means we will not cure diseases. AI is not going to do bad things it will make us live to 300 or forever.

I feel a lot of this video content is essentially saying the importance of technology and why you should be interested in ai, as opposed to how to actually use AI. Am I off base here..

I am still amazed by the sheer number of people who complain about AI. Or those people who say thing like “see, ai was wrong”. The majority of people don’t get it.

if you have 10 workers in your company, and all of the have to do 10 tasks, and AI can easily do 8 of those 10 tasks, you can fire 8 men, and the remaining 2 will do the same work. Work is the sum of multiple tasks. If you can delete many tasks, all the work is simplified, and you need less workers and less time to do the same amount of work. And you can also say to the people that work for you "I am a kind employer, you can continue to work using AI. Obviously, you job will be a lot easier, so you can't expect to be paid the same salary. Honestly, almost all the work is done by AI, I should pay her, not you".

i am 50. so i hope AI solve aging very very soon.

That UBI take at the end is a little weird. Completely disagree with that, but otherwise I agree 100%.

I guess the plaster on your nose is marketing? You trying to set yourself apart or something?

641likes and 37 dislikes

I think the ultimate goal is too end up with AHI Artificial human intelligence

The problem is not that we have to worry about AI deciding to kill us, the problem is that an AI will be developed with that purpose, which is much more likely to happen soon. Our only hope is a more powerful AI that keeps that from happening. This is what is being worked on now, AI to control other AI's. AGI has already happened, and it is not dangerous. You don't need AGI to create an AI that kills.