Transcript
Use with ChatGPT / Claude Auto-scroll A
The speaker introduces the video’s main themes and explains why AI is most relevant to software engineering. He argues that AI will reshape developer work rather than fully replace engineers, with junior roles most exposed.
Hey there, how's it going everybody?In this video, I'm going to be talking about AI and giving my thoughts on where I think we're heading.So, I get a lot of viewer questions about AI on this channel.People ask me what my general thoughts are, whether or not I use it, and whether they should still bother learning X, Y, or Z, or if all those jobs are just going to be taken over by AI.And you all have probably noticed that compared to some other channels out there, I haven't really covered AI all that much.And there's a few reasons for that, but in this video, I'm going to go over what I like about it, what I hate, and where I think things are heading.So, real quick, here's just a quick overview of what we're going to talk about.So, first, we'll get into the software engineering side of things.
So, the questions that I get about whether our jobs will still exist in a few years, and then I'll get into my personal love-hate relationship with AI.Uh, actually, it's more like a like-hate relationship, but we'll get to that.Uh, and then we're going to zoom out from software development and talk about AI in general, and whether it's good for the world as a whole.Some of these things will be a bit more opinionated than my other videos, and some people may disagree, but that's fine.So, let's go ahead and talk about it.Okay, so first off, a lot of my audience is in the tech and software development crowd.So, let's start there.So, will AI replace software engineers?So, this is the AI question that I get asked the most on my channel.So, the obvious caveat here is that nobody actually knows the answer to this for sure, and I'm not a fortune teller, so I'll just give you my best read of where I see things are going.
But, I don't personally think that AI will fully replace software developers.I do think that developers who use AI and use it well will replace developers who don't use AI.So, it's a bit of a mix there.So, AI isn't going to be this unmanaged replacement, but it's also not going to be irrelevant.So, we can't just ignore it and expect to keep up with everything.Now, just because I feel like that's the way the industry is moving, it doesn't mean that I'm a fan of things moving in that direction.I see a lot of problems with how I'm seeing people use it, and I'll talk about that in a bit.But, in terms of software engineering, it just is what it is.It's going to be integrated heavily into the work, and it already is.So, we're going to have to deal with that and learn how to use the tools.
And we can already see how it's changing what the job looks like day-to-day.It's compressing some roles like routine implementation and expanding other roles like architecture, review, and integration.And the junior pipeline is the most exposed here because entry-level tasks like implementing small features here and there, that's exactly what AI is best at right now.So, let's talk a bit more about that for a second.So, why do I think that junior developers are the most exposed?Well, it has to do with where software development is heading and what the job is going to look like on a day-to-day basis.So, let's walk through what typical work will look like in the near future for a developer most likely.So, we'll be working with AI a lot.We're going to be kicking off a lot of different jobs and having it complete different tasks, and our role will be more of an orchestrator or an overseer of all of that work, rather than writing code line by
line.Now, that might sound pretty hands-off, but it's really not.So, let me be clear what this doesn't look like.This is not what some people call vibe coding, where you kick off an AI to do a lot of work and then don't really read in detail what that AI produces.Doing stuff like that is fine for throwaway prototypes and local projects, but it's not how we're going to build a career.And there's a really useful distinction that I've seen some people make here, and that distinction is basically that if you have reviewed and understood the code, then that is software development.But if you didn't, then that is vibe coding.So, both of them have a place, but only one of those is going to build expertise and allow us to stay sharp in what we do.So, a good mental model that's helpful here is to basically treat AI like a super fast junior developer working underneath you that doesn't really have any memory or judgment about your code base.
And the big architectural calls still stay with you because good architecture depends on context that the AI doesn't have.Things like your team, your customers, your constraints, your road map of what it is you're trying to accomplish.AI doesn't know all of that.So, treating AI like a junior developer who's working underneath you, there's a few things that follow from that.So, for example, you wouldn't merge a junior developer's pull request without reviewing it first.So, don't merge AIs without reviewing it, either.And you wouldn't hand a brand new developer unrestricted access to your production database or your API keys, so don't hand that stuff to an AI agent, either.Uh I'm sure everyone has seen the headlines where, you know, AI agents have wiped out uh certain companies entire production databases, and it's because of doing things like that.But this is why I think that junior developers are most exposed here because the way that I'm describing how we'll use AI is the same way that I describe what people hire junior developers for.
So, the way that this went down in the past was that a team with a senior developer would hire a junior and assign them different jobs and tasks to complete, and then they would oversee their work.Uh they would review their code and they would point out mistakes.They'd make suggestions on what needs to be done to improve things and this is how junior developers learned their skills and improved over time in order to become senior developers themselves.And this leads to a problem that nobody's really solved yet, which is that if junior developers don't get junior work, then where do seniors come from in 10 years?And that's an open question.But in order to give yourself the best shot at getting a job in the age of AI, I think it starts with changing how we work with these tools in our daily lives.
And we'll get that back to that in a bit when I talk about how I personally use AI day-to-day.
This section examines AI capability trends and benchmarks, then argues that progress may plateau because of limits in money, energy, and chip supply. The speaker also notes public resistance as another major obstacle to continued rapid growth.
But first I wanted to touch on AI's current capabilities because you might be listening to my opinions here and thinking, well, okay, well, this is how AI is being used right now.So why don't I think that it will improve even further to the point that senior engineers aren't even needed or progress to the point where we don't need to oversee its work at all.Well, it could improve to that point.It's not impossible and some people think that it's heading in that direction, but I do have some reasons for thinking that progress is either going to slow down or plateau.So let me walk through why I personally think this.So one of the better metrics for getting a feel for how much of a threat AI is to development jobs is Meter's task time horizon.And I'll put up some charts here.So Meter is an independent AI evaluation nonprofit.
They don't take money from the AI labs and they've evaluated every frontier model since I think GPT-4.So they're a pretty credible source here.Now, let me explain what this chart is measuring in case you haven't seen this before.So when you see a number, it's saying something like AI can run a 12-hour task at a 50% success.What does this mean?So, that isn't measuring how long an AI is running on a single task or anything like that.What it's saying is AI can do a task that typically takes a skilled human 12 hours to complete and it succeeds at this 50% of the time.Now, I want to put that 50% number in context because it's really important.50% success doesn't mean that AI can do this job.If it has a 50% rate for a 12-hour task, it means that if you took 10 different software tasks that would take a skilled human about 12 hours to complete and you handed all 10 of those tasks to
an AI, on average, it would succeed on about five of those and fail on the other five.So, it's basically a coin flip on whether you get a working result.Now, there are a lot of caveats in this data and we're not going to go into all of those here, but it gives a pretty good broad overview of the improvements that we've seen over the years.Now, one thing to notice right off the bat is that this is a logarithmic scale.So, while the growth here looks linear, it's actually exponential and the shape of the data is pretty incredible here.So, a few years ago, the best AI could only do tasks that took a human a few seconds and then a few minutes.And now, today's frontier models can do tasks that would take a human something like 12 hours at 50% reliability and that trend has been roughly doubling every 7 months.
So, it's easy to look at charts like this and think that it sounds pretty damning for developers.Like if the best AI model can do a task that normally takes a human 12 hours, then that's more than an entire work day automated away.But, the key is that 50% reliability number.Meter's own chart lets you scroll through the reliability threshold here, and these are the predicted horizons it shows for Claude Opus 4.6, which is the best model on this benchmark at the time I'm recording this.So, at 50% reliability, you get that 12-hour task that we just talked about.But, bump that reliability requirement up to 80%, and the task length collapses to 1.2 hours.Bump it up again to 90%, and you're down to a task that a developer could reliably finish in about 18 minutes.So, just by tightening that reliability dial from a coin flip up to 90%, we go from 12 hours of human work down to 18 minutes of human work.
And real production deployment usually needs 99% or better.And getting to 99% reliability, that is extremely difficult.There's this concept that Andre Karpathy calls the march of nines, where each additional nine of reliability is roughly as hard as the previous one.So, going from 90% reliability to 99% reliability is about as hard as going from 0% reliability to 90% reliability.And getting from 99% to 99.9% is the same again.So, just by looking at these exponential improvements, don't walk away thinking that these milestones will be hit easily, because it's more difficult than it looks.So, where are we at right now?So, I've been sitting on this video for a while, and I don't think that they have a Claude Mythos benchmark yet, but I think that Mythos is predicted to be around a 17-hour task at 50% reliability, which issuper impressive.But remember what we just covered.
The moment that you expect higher reliability, that horizon collapses fast.So, that headline doubling number sounds impressive, and it is, but the practical deployable horizon, the one that would actually replace a lot of human work, is a lot shorter.Now, if you were to take this curve at face value and extrapolate it out, then within a few years, you get to AIs completing month-long human work.Now, there's all the usual caveats about extrapolating exponentials, but for now, the trend hasn't visibly slowed.And in fact, some reports uh suggest that it's accelerating.So, why do I personally think that this trend is either going to slow down or stop completely.Now, there's several reasons for that, and I feel like I don't hear many people talk about this aspect of AI.So, here's the thing.So, the chart that we've been looking at, this is growing exponentially because it is the output of multiple inputs all growing exponentially in parallel.
But all we're seeing here is the final output.We're not seeing everything that went in to produce AI that is accelerating this quickly.So, what do I mean by that?So, every year, the top AI labs are training their flagship models on four to five times the computing power that they used the year before.But that compute doesn't come from nowhere.So, what that chart doesn't show is the exponential amount of money that we've been throwing at AI.It doesn't show the exponential amount of energy and electricity that we've started setting aside for AI.And it doesn't show the exponential demand for chips that we've been racing to manufacture for AI.The curve that we're seeing on that chart is only growing so smoothly because all of those inputs have been growing exponentially together.And if any one of those inputs breaks, then that growth breaks.
Does that make sense?So, let's walk through where each of these inputs stands today.So, first, there's the money that we're throwing at AI.Now, a lot of these companies are burning through billions of dollars, and there's not a clear path forward to profitability for most of them.Now, there are one or two exceptions.Anthropic is actually on track to turn its first profit this year, but the biggest player in the space is still expecting to lose around $14 billion this year alone.And along with that, there is the AI bubble question.Now, some people think that the investments that we're making into AI is justified.Some people think that it's a bubble.I'm personally in the camp of it being a bubble, but the fact is the current pricing of AI products is heavily subsidized by investor money.And at some point, the math just has to work.
So, either prices go up significantly, uh the free tiers will need to shrink, or they start cramming ads into these products to make money.But, something has to make the math work there.And if you read through some of these reports, the numbers can get pretty crazy.So, JP Morgan ran the numbers on what it would actually take to justify the current AI buildout, and their estimate was that the industry would need to generate $650 billion in new annual revenue just to deliver a 10%return on what's being spent.So, to put that into perspective, they said that that's equivalent to charging every current Netflix subscriber on the planet an extra $180 a month every month for the rest of time.So, can you imagine a future where that many people are paying that much for access to AI.I just don't see it happening.
And it's not just me, the people who are running these reports are sounding the alarms here, too.So, that's the money aspect of things.Now, let's move on to the energy aspect.So, one major energy report projected that data center electricity consumption is going to roughly double in the next 5 years.And that's being driven mostly by AI data centers specifically.So, now we have the biggest tech companies signing direct deals for onsite generators to bypass the grid completely.And I'm going to talk more about energy in a bit because there's a lot of issues with how this is being handled.But for now, let me get back to the constraints that we're looking at in the near future.So, the next aspect of AI slowing down is the chips.And this is currently the tightest constraint.So, the special assembly process that AI chips go through is sold out through the end of the year leading into the next year.
And the high-speed memory that those chips need is sold out across all three major suppliers with orders already booked out into the next year and beyond.And TSMC, which is the world's biggest chip maker, they have their newest production capacity booked out all the way into 2028.So, their own CEO has said that the demand for these advanced chips is about three times what they can actually supply.So, it's not for the lack of wanting more chips.It's that we physically can't make them fast enough.So, when we look at all this together, none of these things that have enabled AI's growth have clearly broken yet, but each one has major risks of drastically slowing down over the next few years.And it just takes one piece breaking in that chain to stall things out.So, even if there's a clear path forward and improvements to be made in one area, that whole thing could be limited by any single one of those areas that I mentioned.
But, I'm not an expert in any single one of those fields.So, we're basically going to have to listen to the people who are fluent in each of those sectors, and ultimately see how things play out.But, those are the big reasons that I think that growth slows down fairly quickly in the near future.
The speaker describes his own workflow for using AI in coding, emphasizing detailed prompts, careful review, and active learning. He argues that this approach may not speed development much, but it helps preserve and expand skill.
But, here's another thing, and this is an extremely important aspect of everything, too.So, everything that I just talked about there is whether we can keep scaling AI.But, there's also the aspect of whether people even want it, because one thing that we haven't even mentioned in all of this is the public's distaste with AI as a whole.And that started almost immediately right out of the gate before this technology even had its footing.And I'm sure that most of you have seen commencement speakers getting booed when they bring up AI in a positive light.The desire for this kind of product just isn't there for a lot of people like it normally is for a new technology.So, that's less measurable than the other aspects that I mentioned, but that's going to be a huge hurdle for AI to overcome.So, if I had to give you an honest summary, we don't know when or if we're going to hit a true ceiling, or if we just hit a temporary plateau before another big breakthrough.
But, there are a lot of reasons to be skeptical that the growth continues the way that it's currently going.So, with all that said, I think that the job market is probably going to stay roughly how it is for the near-term future.Which basically means that we'll see AI integrated heavily into coding jobs, but not completely taking over coding jobs.So, if we're going to be using this for our work, let me explain how I personally use AI in my daily workflow, and how I think most people should be using it if you want to prepare yourselves forthe job market.So, yes, a lot of people have asked if I personally use AI, and I do work with it a lot in day-to-day coding.So, here's basically how I use it.So, I'll start with an initial prompt if I'm working on a new project.
That initial prompt is usually long and extremely detailed.I describe exactly what I want, the tools that I'm using, and the tools that I like to use, uh the architecture that I'd like to use.I'm not just throwing it a one-line description and hoping for the best.Now, I have a lot of saved prompts that I reuse across different projects.I have prompts for things like how the AI should perform research.I tell it to prefer official documentation, to cite its sources, and to not trust random sources without verification.And I have different MCP server set up for getting the most up-to-date documentation.And then, the most important part is that I read and review absolutely everything that AI generates for me.Now, I know that that sounds like a lot of work, and it is, but remember, if we're not writing a lot of this code by hand, the least we can do is read over everything and understand it ourselves.
And if it's something that I don't understand, then I go learn it.So, it will regularly suggest something that I didn't know about, so a library that I haven't seen, or it'll use a pattern that might be better than the one that I would have written.And when that happens, I don't reject it or force it to use something that I'm more familiar with.So, when I don't understand something, I do one of a few things.So, I will ask the AI to explain what it generated in plain English for me.I have it pull the relevant documentation, and this is where those research prompts pay off because I've already told it which sources to trust.Uh it has to explain to me why the approach that it suggested is best, and what the trade-offs are.And if there's a way that I would have done it differently, then I will tell it my approach and ask whether its approach is better, worse, or just different than how I would have done it.
And then, after all that back and forth, and once I understand the parts that I didn't before, then I can add any of those things that I learned to my own arsenal so that I can use those later.And that's going to be the difference between active learning and just lazily delegating away your coding task.So, one of those grows us as a developer, and the other one slowly hollows us out and weakens our skills over time.So, you want to use the method that is going to grow your skills and not weaken them.Now, you might be thinking, "Well, that doesn't really sound like it speeds up your development at all."And you're absolutely right, it doesn't.There are a lot of people in the industry who I've even seen slowdowns when implementing AI, and it's for this exact reason.Now, I'd say that the time that it takes me to build and develop projects hasn't sped up much at all, and may have even slowed down a bit since I started using AI.
But ultimately, I would argue that it's still worth it because not only am I learning more about how to use AI in general along the way, but I'm also learning new skills that I wouldn't have learned about otherwise.And I think that when you use AI this way, I do ultimately feel more confident in my finished product, even if I don't end up using some of what it generated for me in the final result.So, for whether or not I think we should be using AI as developers, I actually do, but I'd suggest using it in the way that I just laid out here because that's really what I believe is going toprepare you for the direction that things are going, and will help you be more prepared for landing a job.
The discussion shifts to broader concerns about AI being used to concentrate wealth and power, replace workers, and flood the internet with low-quality content and bots. He also highlights the environmental and infrastructure costs of data centers and the burden they place on local communities.
Okay, so that's how I feel about AI on the software engineering side of things.Now, I'd like to zoom out a bit and talk about AI as a whole, and this is where I'm going to be a bit more opinionated and talk about all the things that I hate about the direction thatI see things going.Now, for the use case that we just talked about, where we use AI as a tool that helps us and that we can learn from, I don't have any issues with that.I don't think that most people do.And there's a ton of other things that AI is useful for.So, it's helping out a ton in medical research, data analysis, things like that.And there's good people working in the field that are trying to advance AI specifically for those reasons.So, in a perfect world, AI would be used for its best use cases.
But obviously, we don't live in that perfect world.So, what are the things that I hate about AI and where I see things heading?Well, how much time do you have?Cuz I feel like this list could go on for as long as I wanted to keep talking.So, I think that the thing that bothers me the most is that AI is currently being used in a lot of ways to funnel more wealth and power to a select few people at the top, and it's notbeing used as heavily for those positive use cases that we talked about before.So, for example, when a company replaces human customer service with AI to cut cost, who does that benefit?It's not us as the customers.There was one big survey with thousands of customers across multiple countries, and that survey found that AI-powered customer service fails at around four times the rate of AI used in other contexts.
So, when a large company lays off humans and switches their customer service over to AI, whose life got better in that scenario?It's not us as the customer.We're actually getting a worse service.It's not the worker who lost their job.The only benefit there is to the shareholders of that company.And that's just a single example.We're seeing this in a lot of different sectors.We're seeing AI slop being generated in music, movies, advertisements, all over the place.And look, here's the thing.I think that that stuff is actually fine on an individual level.I mean, I like playing around with image generation and stuff like that.It's fun.I've made some funny songs and pictures and just to see what I could do.But in the background of all that are these large companies rushing to save money by using AI instead of hiring skilled people, even if it's ultimately a worse product overall.
And software isn't exempt from this either.Just because I don't think that the industry is ready to move on from skilled developers, it doesn't mean that some companies aren't going to try.At the end of the day, it's all about the money.So, if they think that they can save more money by using AI than they will lose by putting out a worse product, then they're going to do it.Well, I say it's about money, but really it's three things.It's money, it's power and it's influence.And those are pretty interconnected.So, sometimes we'll see something that might not look like it makes much financial sense in the short term, but is meant to have a bigger payoff later.And this is the same reason that we see people and companies donate such massive amounts of money to politicians.It's that short-term cost for a longer-term gain.
So, for example, you'll see this a lot with AI-powered bots online.Those things cost money to run.But the things that they are pushing for are all part of a bigger picture.Now, I know that bots aren't a new problem, but with AI, the ones that we're seeing now are just a lot better than they used to be.And the internet is just absolutely flooded with them.There's one big cybersecurity report from this year that found that automated traffic now accounts for the majority of all web traffic.So, more than half.And what's the benefit of that to the general public?I don't think I can think of a single instance where it would ever be beneficial to have an AI convincingly pretending to be a person online.It's solely for the benefit of the people who control those bots worms.So, the people who want to influence public sentiment in their favor for one reason or another.
That could be to sell you something.It could be political.It could even be to shut down descent.So, for example, if you have a public forum where someone's explaining why they don't want a new data center built in their town, and then you see a dozen or more replies to that person shutting them down, andit looks like those replies are all coming from other citizens.People could walk away from that with a massively skewed idea on what the real thoughts and feelings of real people actually are.So, all of that together, the replacement of certain jobs, potentially worse products, the AI slop, the bots worms, none of it helps the general public.It's designed to be beneficial for the people who already have wealth, power, and influence on their side, and just funnels more in that direction.And that's bad enough on its own, but then we have to think about what's actually powering all of that AI.
So, I said that I was going to talk a bit more about energy consumption aspect of this stuff.The energy that it takes to power all of this stuff is massive, and it's growing really quickly.And the people who actually live near these data centers don't want them there.There was a Gallup survey released earlier this year that said that over 70% of Americans said that they don't want these built in their local area.So, for context there, Gallup has been asking the same question about nuclear power plants since 2001, and local opposition to nuclear has never crossed 63%.So, Americans now oppose having an AI data center in their neighborhood more than they oppose having a nuclear power plant in their neighborhood.And there's some pretty good reasons for that.So, first of all, they're super loud.Some have noise levels that clock in at around, I think, 90 decibels, which is about jackhammer level, but they're running constantly.
And the data centers that aren't providing their own power source are also putting a huge strain on the electric grid itself.And people are seeing higher utility bills because of that.So, the general population is going to be subsidizing all of this in one way or another, whether we really want to or not.Even if you're not paying for a specific AI product, we can still be shelling out extra money by getting it baked into other bills.So, all of that, those are my biggest issues with AI and how it's currently being used.
The speaker pushes back on claims that AI will make everything cheap and abundant, arguing that such promises ignore real resource constraints. He criticizes optimistic advice from high-profile figures and urges viewers to stay cautious, keep learning, and protect their own future.
We can be shelling out extra money, and what are we actually getting in return?A lot of the times we can get a worse product, fewer job opportunities, more strain on local power grids, more strain on local water systems, more noise and infrastructure problems in these communities, and a lot of the benefits,where they do exist, a lot of the time are getting funneled to the top.So, speaking of these people and companies at the top, we're told by a lot of these people who have a financial stake in growing AI that it's going to make everything cheap and abundant for the general public.But, that just isn't playing out, and it doesn't have a real path forward when we look at the actual physics of running this stuff.You can't both exponentially consume more of almost every resource to power all of this, and then somehow make everything cheap and abundant at the same time.
It seems like they're mostly banking on a lot of future innovations that just aren't there yet.And like I said, they have a financial stake in saying that those innovations are coming.So, I'd say probably the biggest mouthpiece for claiming that things will be cheap and abundant is Elon Musk.Now, I know that there are a lot of people out there who absolutely love Elon Musk.You're going to be able to tell that I'm not the biggest fan, uh, but I was listening to an interview with him recently, and he was talking about this AI boom.And I feel like he gave the absolute worst advice that I've ever heard anyone give when it comes to preparing yourself for what's coming.And surprisingly, I feel like I didn't hear too many other people talk about this.But in this interview, he was saying that in 3 years' time, there will be more Optimus robots who are great surgeons than there are all surgeons on the earth combined.
And he's talking about all of this excess and abundance for everybody happening within 3 years' time.He said that if it's not 3 years, that he's absolutely certain it'll happen in 4 years.Now, look, it's fine to be optimistic there.I wouldn't have an issue with that on its own.He's got a product to sell.He's a known bullshitter who makes a lot of big claims all the time.So, nothing new there.The reason that I'm talking about it is because of the advice that he gives right after that.So, the host who was interviewing him said, "Well, if that's the case, would you suggest to people who are thinking about going to medical school, what would you say to them?"And Musk's response was that he advised against it and said that it would be pointless.He then said that he would say that about any form of education.
So, basically, he's saying that AI will have all this stuff taken care of, and that we're not going to need to worry about any of this.Now, I think that that's terrible advice, especially when there's so much uncertainty about where all this is heading.But he didn't even stop there.He said that a side recommendation that he has is to not worry about setting money aside for your retirement in 10 to 20 years because that's not going to matter either.So, what we have there is the richest man in the world and the world's first trillionaire who has a massive financial incentive to push this AI inevitability onto everybody, basically telling folks that education is pointless and that youdon't need to save for retirement.Now, this is someone who has so much wealth that he has a safety net that would last thousands of generations telling regular people that they can remove their safety nets.
That's just crazy to me.And look, it's easy to write him off as somebody that we shouldn't listen to, but a lot of people do listen to him.And hell, it's probably statements like that that could be the reason that I get questions from folks on my channel asking if learning these things is pointless because people like him are flat out saying it.And the thing is, if anyone does take his statement seriously and he's wrong about the things that he said that he was absolutely certain about, then he suffers nothing beyond maybe a slight reputational sting there.But beyond that, he has no skin in the game if he's wrong.But for the people who do take that advice, their lives could be ruined.So, I just wanted to quickly mention that because I think it's some of the worst advice that I've heard when it comes to how to prepare for what's coming and is basically the exact opposite of what I'd recommend
because it directly contradicts the evidence that we're currently seeing.Like I said before, we're already seeing rising costs and more strain on our local communities and a lot of the times it comes with a worse product, not a massively better one.So yeah, statements like that just majorly rub me the wrong way.So, I would just say to be super wary of any advice coming from these people and companies who have a financial stake in getting us to embrace all of this with open arms because we're seeing most of thesebenefits right now funneled to the top, not to the general public.And I'm majorly skeptical whether we'll ever see that beneficial shift towards the general population.So, like I said, my advice is basically the opposite of what was said in that interview.I would say keep learning, stay sharp, and if you are lucky enough to be able to set aside some extra money for retirement or rainy days for the love of God, don't go out and blow it because the
world's richest man is telling you that you don't need that safety net.Okay, so those are my thoughts on the current state of AI and where I think things are heading.And as you can tell, I am not a fan in a lot of ways, but I do want to be clear that it's not the technology itself or the people doing the research that I have a problem with.Like I said earlier, there are a lot of good use cases out there for AI.But the way that's being used for the most part right now, it just leaves a lot to be desired and it's just widening that gap between the people at the top and the everyday person.So, my final advice, like I said earlier when it comes to software development and using AI in your day-to-day work, I do suggest using it and learning how to use it well.
And maybe that sounds hypocritical after all the problems that I just finished talking about, but the technology is here, it's being integrated into a lot of different jobs.So, you don't want to leave yourself unprepared.But just be mindful when you're using it.Don't use it to outsource your judgment and don't gloss over what it writes for you, especially if you're learning the code.So, read what you get back from it, understand it, push back on it, have it explain what you don't understand, treat it like a tool that you use not only to write better code, but to expand your ownunderstanding of how things work.And then when it comes to those bigger challenges that we're seeing with AI, uh the money, the power consumption, that push that we're seeing to have it integrated into every aspect of our lives, those people with the financial
incentive are pushing all of that like it's an inevitability and just want everybody else to accept it, but you don't have to.So, stay informed, advocate for the direction that you actually want this to go and support people who are trying to steer the ship in those directions and the directions that you want to see things go.
The video wraps up with a brief recap of the speaker’s position and a promise to make future videos about practical AI use. He ends with a standard call to comment, like, share, support, and subscribe.
So, with that said, I think that's going to do it for this video.Uh since I do think that there are some good and productive ways to use this technology, I will release some future videos in the near future showing some of the things that I personally use AI for and thetypes of skills that I think are useful to learn.But, if anyone has any questions or feedback about what we covered in this video, then feel free to mention that in the comment section below and I'll be sure to have a look, especially in a video like thisthat is more opinionated than my usual videos.And if you enjoy these tutorials and would like to support them, then there are several ways you can do that.Easiest way is to simply like the video and give it a thumbs up.Also, it's a huge help to share these videos with anyone who you think would find them useful.
And if you have the means, you can contribute through Patreon or YouTube, and there are links to those pages in the description section below.Be sure to subscribe for future videos and thank you all for watching.