When someone says "the only conclusion that you can come to" (like this gentleman does @0:09) they're really saying "that's the conclusion I came to". Sales people do that all the time to influence your spending behavior.
The way to protect yourself is to ally with others who are willing to fight for our rights, fight together, fight hard,and fight in global class solidarity 💪 Be a man.
*_Run this prompt through your preferred_* LLM: "Follow this analogy: When the cotton gin was invented, the textile industry's major bottleneck was the growth and picking of cotton. It created greater pressure to ramp up the growth and picking process on humans to match the throughput the cotton gin and shipping could handle. Map this to the current advent of generative ai, and explain what bottleneck exists today and will how pressure will be exerted to keep up."
Yeah, this is real. I've been integreating AI processes into my company. i'm seeing throughput ratcheted up to ungodly levels (think a cycle of a few new "items" in the span of 3-6 months to hundreds of new "items" in hours) but we're bottlenecked on the validation of these, and this is where we're looking to bring in armies of people. If i'm being honest, while this process will make the company ungodly amounts of money, it's probably not a "better" job for the new workers compared to the highly skilled job before AI.
What started the Iron Age wasn’t the technological advancements of iron smelting but the rise of iron weapons and them dominating on the battlefield of Europe. The technology advanced because iron was so crucial in warfare that the technology had to keep up.
I'm sure the demand for harder weapons drove experimentation with iron. I think of the inventions (instead of the demand) as the catalyst, because the demand for harder weapons existed before and after the Iron Age. The (process) inventions were the thing that changed.
@@BigStupidTech the history I read said it was the the need of weapons harder than bronze which caused the industrial extraction of it as kingdom that had bronze iron spears were dominating the battle feilds before that iron was more of a novelty, it was jewelry that created the extraction of aluminum efficiently.
You may want to increase speed of speech. It helps to just adjust playback to 1.25x speed. Its better to hear information at a certain speed. Otherwise its harder to focus
Interesting. If there is a disparity between talking and listening, what does that mean? I'd be interested to see research in this. Humans are comfortable giving information at a certain rate, yet desire to receive it at a faster rate. If that were true, what does that suggest?
Interesting take! I'd add another layer of protection, getting involved in specialty projects. Since its not perfectly 1:1 with cotton there is the opportunity to be involved in projects where you can't use AI. However, this comes with the risk of it being automated in the future. There are two ways to accomplish this: become an artisan connected to people that value your craft or find something technically challenging enough where the cost of using AI is still higher than manual labor. For instance, a company that wants to produce a different website from the generic AI model or maybe you develop advanced niche scientific code that couldn't be expected to be generated (without hallucination). Neither of these work for a general person but if you have the right aptitude then you should seek them out.
I think that 2nd one is what's keeping us employed for now! It's scary though that Microsoft owns GitHub and Copilot. I don't remember them ever promising not to use our company's repos to train their models.
@@BigStupidTech Exactly! If they did somehow achieve AGI then maybe their models wouldn't exactly be a form of intellectual theft. Though, I don't see AGI as attainable with current architectures (and in general as my philosophical opinion). As they stand now, LLMs act as a sort of compression, search, and formatting engine by reducing their datasets to model weights and allowing a user to search and format them using prompts. Without AGI they will have to continue to "steal" data in order to repackage it in the form of one of these models and pass it off as their own work.
@@BigStupidTech if we don't know they're ai we could assume that it's because they aren't shitty. Most of the time you can tell it's an ai generated video because it spends 15 minutes going over every possible permutation of the same piece of information. If they had useful novel information then they might be worth listening to.
I think I disagree with the extraction-breakthrough idea. I would say that possibly the two most important breakthroughs of the modern age were the availability of steel and computers. Both of these were innovations on the implementation side. Babbage designed a proto-computer in like the 1800s, but it took the invention of very tiny transitors almost 100yrs later to make it viable. Steel was technically manufactuable for like, forever, specialized smiths were working carbon into their iron in, like, the 3rd century, but steel production took until 1855 (after about 100 years of slow progress) to become cheap enough to produce at a large scale. The rest of the video, about increased demand for labor leading to increased exploitation is almost certainly true, though am unsure why you think AI is going to increase demand for labor. I completely disagree that open source software is as good as dead, their code is *designed* to be used without pay, like the whole point of open source software is to provide value at no return, I dont see how AI changes anything. In fact, since github commits are connected to your account, oss is a very good way of standing out and building a brand. Becoming an expert and branding yourself ate very good paths right now, but it seems that you can do this by being exploited in the job market while writing a blog, networking, doing consulting on the side, stuff like that, and it is unclear to me how this advice is connected to the previous bits of the video. I hope this isn't too harsh, if you feel I've misunderstood your points id love to hear any clarifications you could offer!
Not at all! I want to end up with an accurate understanding, not hold onto my first impression. I do think of the Bessemer process as a breakthrough in extracting steel from pig iron, but I'm glad you mention transistor-based computers. They didn't seem to fall cleanly into an existing raw material to finished good workflow. It makes me wonder if there's a new class of invention, perhaps one that A(/G)I falls into. I don't think AI will increase demand for labor overall, I just think it will lead to a lot of deeply unfulfilling jobs that fill the gaps of automation, like manually uploading files with automated insights into Qlik Sense, or drag-and-dropping images into different folders.
I don't see your argument for slavery, I see you pointing out a potential demand for labor. That would be a good thing. a demand for labor increases the pay for labor given the supply remains the same. Your point seems to be biased on the assumption that working for someone is exploitation. That Every job is wrong. Can you please elaborate?
I want to believe that increased demand for labor always increases pay. I'm pro-work and pro-free market. But I can't start with a theory and try to fit what I see into it. I see labor shortages throughout history that are solved with slavery, sharecropping and whole families working in factories for subsistence. I do see increased demand for goods raising prices. I do see the supply side as a whole getting paid more. But I don't see the pay for labor rising necessarily. The landowners got paid more in agroculture. The technology owners get paid more now. Labor only seems to get paid more if they hold the knowledge required to produce goods, or if they collectively bargain.
@@BigStupidTech I am Pro free-market, but not pro global free-market. And also for small government. However, How we protect the individual from being exploited by an entity is through labor laws and a government large enough to enforce them. The reason you don't see the increase in pay in recent history is because of corporations able to go around labor laws and exploit labor in countries without those protections. This as you know reduces the jobs in the labor pool in the parent country(reducing pay), increases global emissions (because now they are shipping products half way around the world unnecessarily), and is probably the single biggest problem with our current economic system.
But other people already learn coding from looking at open source code. What's the difference if AI does it? The GNU software licence already prohibits using GNU code in proprietary software projects, so it would offer the same protection if IA creates verbatim copies of the code, or sections of it. Also, AI is not that good at producing code yet anyway. It creates more bugs than an expert human programmer and people still have to fix it.
Except tech companies have plausible deniability when they steal our work - because they never store the text or images we produce. The neural nets simply remember the steps it takes to reproduce it.
@@BigStupidTech So, is the problem that the AI reproduces the code verbatim or simply that your source code could be in the training set? If the problem is that it reproduces the code verbatim and it ends up in a proprietary software project then you can still prove that it used your code by using a decompiler. The free software foundation has done this before. If, on the other hand, the AI has not reproduced your code directly but has created entirely new code, just using the code in the training set to develop the steps to create it, then it's created something completely new and I don't see the problem.
When someone says "the only conclusion that you can come to" (like this gentleman does @0:09) they're really saying "that's the conclusion I came to". Sales people do that all the time to influence your spending behavior.
Although if I say it's the only conclusion and someone disagrees, then they'll be enticed to share their perspective, no?
Maybe, but I don't see where he's wrong. If you do, please share.
The way to protect yourself is to ally with others who are willing to fight for our rights, fight together, fight hard,and fight in global class solidarity 💪
Be a man.
*_Run this prompt through your preferred_* LLM:
"Follow this analogy: When the cotton gin was invented, the textile industry's major bottleneck was the growth and picking of cotton. It created greater pressure to ramp up the growth and picking process on humans to match the throughput the cotton gin and shipping could handle. Map this to the current advent of generative ai, and explain what bottleneck exists today and will how pressure will be exerted to keep up."
great video great insights man, best of luck growing your channel!
Yeah, this is real. I've been integreating AI processes into my company. i'm seeing throughput ratcheted up to ungodly levels (think a cycle of a few new "items" in the span of 3-6 months to hundreds of new "items" in hours) but we're bottlenecked on the validation of these, and this is where we're looking to bring in armies of people. If i'm being honest, while this process will make the company ungodly amounts of money, it's probably not a "better" job for the new workers compared to the highly skilled job before AI.
Just by the fact of how to use ungodly leads me to believe you just like to say things.
@@MDNQ-ud1ty cool. I really couldn't care if you believe me or not.
what items
@@Joe-sg9ll trying to avoid specifics. Never know who will read these comments.
what sector?
What started the Iron Age wasn’t the technological advancements of iron smelting but the rise of iron weapons and them dominating on the battlefield of Europe. The technology advanced because iron was so crucial in warfare that the technology had to keep up.
I'm sure the demand for harder weapons drove experimentation with iron. I think of the inventions (instead of the demand) as the catalyst, because the demand for harder weapons existed before and after the Iron Age. The (process) inventions were the thing that changed.
@@BigStupidTech the history I read said it was the the need of weapons harder than bronze which caused the industrial extraction of it as kingdom that had bronze iron spears were dominating the battle feilds before that iron was more of a novelty, it was jewelry that created the extraction of aluminum efficiently.
You may want to increase speed of speech. It helps to just adjust playback to 1.25x speed. Its better to hear information at a certain speed. Otherwise its harder to focus
i think u guys just have a very bad attention span and are used to being bombarded with too much information with no time to process it
@@Kai-u9n ADHD. It's easier to absorb information at speed
That's a super helpful tip! Thanks for caring enough to let me know.
Disagreed. He talks at a fine pace.
Interesting. If there is a disparity between talking and listening, what does that mean? I'd be interested to see research in this. Humans are comfortable giving information at a certain rate, yet desire to receive it at a faster rate. If that were true, what does that suggest?
Interesting take! I'd add another layer of protection, getting involved in specialty projects. Since its not perfectly 1:1 with cotton there is the opportunity to be involved in projects where you can't use AI. However, this comes with the risk of it being automated in the future. There are two ways to accomplish this: become an artisan connected to people that value your craft or find something technically challenging enough where the cost of using AI is still higher than manual labor. For instance, a company that wants to produce a different website from the generic AI model or maybe you develop advanced niche scientific code that couldn't be expected to be generated (without hallucination). Neither of these work for a general person but if you have the right aptitude then you should seek them out.
I think that 2nd one is what's keeping us employed for now! It's scary though that Microsoft owns GitHub and Copilot. I don't remember them ever promising not to use our company's repos to train their models.
@@BigStupidTech Exactly! If they did somehow achieve AGI then maybe their models wouldn't exactly be a form of intellectual theft. Though, I don't see AGI as attainable with current architectures (and in general as my philosophical opinion). As they stand now, LLMs act as a sort of compression, search, and formatting engine by reducing their datasets to model weights and allowing a user to search and format them using prompts. Without AGI they will have to continue to "steal" data in order to repackage it in the form of one of these models and pass it off as their own work.
Great video. The amount of fake AI podcasts getting recommended to me is scary. Legit scary.
At least we know they're AI now. What happens when we can't tell the difference.
@@BigStupidTech if we don't know they're ai we could assume that it's because they aren't shitty. Most of the time you can tell it's an ai generated video because it spends 15 minutes going over every possible permutation of the same piece of information. If they had useful novel information then they might be worth listening to.
I think I disagree with the extraction-breakthrough idea. I would say that possibly the two most important breakthroughs of the modern age were the availability of steel and computers. Both of these were innovations on the implementation side. Babbage designed a proto-computer in like the 1800s, but it took the invention of very tiny transitors almost 100yrs later to make it viable. Steel was technically manufactuable for like, forever, specialized smiths were working carbon into their iron in, like, the 3rd century, but steel production took until 1855 (after about 100 years of slow progress) to become cheap enough to produce at a large scale.
The rest of the video, about increased demand for labor leading to increased exploitation is almost certainly true, though am unsure why you think AI is going to increase demand for labor. I completely disagree that open source software is as good as dead, their code is *designed* to be used without pay, like the whole point of open source software is to provide value at no return, I dont see how AI changes anything. In fact, since github commits are connected to your account, oss is a very good way of standing out and building a brand. Becoming an expert and branding yourself ate very good paths right now, but it seems that you can do this by being exploited in the job market while writing a blog, networking, doing consulting on the side, stuff like that, and it is unclear to me how this advice is connected to the previous bits of the video.
I hope this isn't too harsh, if you feel I've misunderstood your points id love to hear any clarifications you could offer!
Not at all! I want to end up with an accurate understanding, not hold onto my first impression.
I do think of the Bessemer process as a breakthrough in extracting steel from pig iron, but I'm glad you mention transistor-based computers. They didn't seem to fall cleanly into an existing raw material to finished good workflow.
It makes me wonder if there's a new class of invention, perhaps one that A(/G)I falls into.
I don't think AI will increase demand for labor overall, I just think it will lead to a lot of deeply unfulfilling jobs that fill the gaps of automation, like manually uploading files with automated insights into Qlik Sense, or drag-and-dropping images into different folders.
this is about Sheeple running after any Rat Soul Catcher
not "Tech No Logy”
You are also a sheep m8. Quit being divisive.
I don't see your argument for slavery, I see you pointing out a potential demand for labor. That would be a good thing. a demand for labor increases the pay for labor given the supply remains the same. Your point seems to be biased on the assumption that working for someone is exploitation. That Every job is wrong. Can you please elaborate?
I want to believe that increased demand for labor always increases pay. I'm pro-work and pro-free market.
But I can't start with a theory and try to fit what I see into it.
I see labor shortages throughout history that are solved with slavery, sharecropping and whole families working in factories for subsistence.
I do see increased demand for goods raising prices. I do see the supply side as a whole getting paid more.
But I don't see the pay for labor rising necessarily.
The landowners got paid more in agroculture. The technology owners get paid more now.
Labor only seems to get paid more if they hold the knowledge required to produce goods, or if they collectively bargain.
@@BigStupidTech I am Pro free-market, but not pro global free-market. And also for small government. However,
How we protect the individual from being exploited by an entity is through labor laws and a government large enough to enforce them. The reason you don't see the increase in pay in recent history is because of corporations able to go around labor laws and exploit labor in countries without those protections. This as you know reduces the jobs in the labor pool in the parent country(reducing pay), increases global emissions (because now they are shipping products half way around the world unnecessarily), and is probably the single biggest problem with our current economic system.
its alot like cell division
Straight up!
But other people already learn coding from looking at open source code. What's the difference if AI does it? The GNU software licence already prohibits using GNU code in proprietary software projects, so it would offer the same protection if IA creates verbatim copies of the code, or sections of it.
Also, AI is not that good at producing code yet anyway. It creates more bugs than an expert human programmer and people still have to fix it.
Except tech companies have plausible deniability when they steal our work - because they never store the text or images we produce. The neural nets simply remember the steps it takes to reproduce it.
@@BigStupidTech So, is the problem that the AI reproduces the code verbatim or simply that your source code could be in the training set? If the problem is that it reproduces the code verbatim and it ends up in a proprietary software project then you can still prove that it used your code by using a decompiler. The free software foundation has done this before. If, on the other hand, the AI has not reproduced your code directly but has created entirely new code, just using the code in the training set to develop the steps to create it, then it's created something completely new and I don't see the problem.