Latest / DevQuestions with Tim Corey / 299. Hope - Why Software Developers Should Be Hopeful

Transcript

299. Hope - Why Software Developers Should Be Hopeful · DevQuestions with Tim Corey

• 0:00
  SPEAKER_01: Right now, in the face of so many AI announcements, it may seem hopeless.
• 0:04
  SPEAKER_01: The job market is soft.
• 0:06
  SPEAKER_01: Employers are cutting back in the name of AI efficiencies.
• 0:10
  SPEAKER_01: Good developers are struggling to get jobs, and we're being told that AI is only going to further press into the job market.
• 0:18
  SPEAKER_01: So, how do we stand up against this onslaught?
• 0:21
  SPEAKER_01: Is there any hope?
• 0:22
  SPEAKER_01: Or is AGI going to arrive and take away all of our jobs and leave us in poverty?
• 0:27
  SPEAKER_01: Now I'm going to give the answer right up front, and then I'm going to back it up with lots of data.
• 0:32
  SPEAKER_01: So the answer is that we absolutely have a bright future ahead of us.
• 0:37
  SPEAKER_01: In fact, as developers, we're going to have a ton of work to do in the coming years.
• 0:42
  SPEAKER_01: So much so that I'm really concerned we don't have nearly enough qualified developers who are actually prepared to do the work.
• 0:51
  SPEAKER_01: But more on that in a bit.
• 0:53
  SPEAKER_01: We do have a bright future ahead, but we're currently hitting a bit of a speed bump.
• 0:57
  SPEAKER_01: Let's talk about the true realities of where we're at and separate facts from fears in this episode of Dev Questions.
• 1:14
  SPEAKER_00: Welcome to the Dev Questions Podcast with Tim Corey.
• 1:17
  SPEAKER_01: Software development is more than just writing code.
• 1:20
  SPEAKER_01: So let's talk about the rest of it.
• 1:22
  SPEAKER_01: Specifically, let's talk about the real and growing hope around software development.
• 1:27
  SPEAKER_01: And we're going to start off with where we are and in some ways tears down some ideas of what people think about this AI revolution.
• 1:37
  SPEAKER_01: Now, to do this, I'm actually going to do something I don't normally do in-depth questions.
• 1:42
  SPEAKER_01: And that is, I'm going to switch over and show you article after article after article.
• 1:47
  SPEAKER_01: So if you're listening to the podcast, no worries.
• 1:50
  SPEAKER_01: I'm going to talk you through it.
• 1:51
  SPEAKER_01: But if you're watching on YouTube, then you get to see exactly what I'm talking about.
• 1:56
  SPEAKER_01: Okay, so let's jump over and start looking at these.
• 1:59
  SPEAKER_01: And I've pulled out various headlines, and then we're going to do is look at the story behind the headline because too often people see headlines and they stop there.
• 2:10
  SPEAKER_01: So this first one is when I came across, it's just a good starting point to talk about this idea that there's a lot of hype around AI, and there's a lot of headlines that are potentially true, but not necessarily the way you think they are.
• 2:27
  SPEAKER_01: So what they're saying can be true while at the same time not be what it sounds like.
• 2:33
  SPEAKER_01: So this first one, Google engineer says Claude Code built in one hour what her team spent a year on.
• 2:40
  SPEAKER_01: Now, if you think about that, a Google engineer just said, first of all, Claude, what's interesting that they use Claude, not Gemini, but that Claude Code did something so amazing that basically wiped out a year's worth of work in just one hour of Claude working on it.
• 2:59
  SPEAKER_01: Now, if you hear this, you think software development is dead.
• 3:02
  SPEAKER_01: Why would you pay a Google engineer or a team of engineers for a year's worth of work when Claude can do it in one hour?
• 3:10
  SPEAKER_01: That would be a great question to ask.
• 3:12
  SPEAKER_01: But let's dive a bit deeper because there's an update to the story.
• 3:16
  SPEAKER_01: So the update as of January 4th of this year is that while she wants to cut through the noise because her comment attracted a bunch of attention, but she says Google has built several versions of the system over the past year.
• 3:32
  SPEAKER_01: There are trade-offs and there hasn't been a clear winner.
• 3:34
  SPEAKER_01: When prompted with the best ideas that survived, okay, so here's the thing: they learned a bunch of things by trying different things, and then they took the best of those and used those as prompts.
• 3:46
  SPEAKER_01: So they've done all the work already and used that work to give a better prompt to Claude.
• 3:51
  SPEAKER_01: So that's a starting point.
• 3:53
  SPEAKER_01: When prompted with the best ideas that survived, coding agents can generate a decent toy version in about an hour.
• 4:02
  SPEAKER_01: Now, is that the same thing as what the headline said?
• 4:06
  SPEAKER_01: The headline says that Claude built in one hour what her team spent a year on.
• 4:12
  SPEAKER_01: It's not really what happened.
• 4:15
  SPEAKER_01: They built a toy version.
• 4:17
  SPEAKER_01: What I built this weekend isn't production grade, it's a toy version.
• 4:22
  SPEAKER_01: I'm surprised the quality of what's generated in the end because I didn't prompt into depth about design choices yet.
• 4:28
  SPEAKER_01: Claude Co.
• 4:29
  SPEAKER_01: is able to give me some good recommendations.
• 4:32
  SPEAKER_01: So the original article says that one hour her Claude did something that took her team a year.
• 4:43
  SPEAKER_01: But the actual update is, well, it was a toy version.
• 4:47
  SPEAKER_01: It was based upon our best information, and it gave us some good ideas.
• 4:53
  SPEAKER_01: But that's not the same thing as building a production-ready application.
• 4:57
  SPEAKER_01: So the title doesn't match the actual content.
• 5:00
  SPEAKER_01: And this is where I want you to start separating out the hype of something from the reality of it.
• 5:07
  SPEAKER_01: Because too many people are looking at the hype of this and saying, oh, I can start laying off developers.
• 5:15
  SPEAKER_01: And this is true for management, for CEOs, for all the rest, because they they don't want to pay for us.
• 5:22
  SPEAKER_01: And who would?
• 5:23
  SPEAKER_01: Because we're expensive.
• 5:24
  SPEAKER_01: But when they start seeing this, they start thinking this is reality and they don't dig into the specifics.
• 5:31
  SPEAKER_01: By the way, if you have management you want to share us with, that'd be a great thing to do.
• 5:36
  SPEAKER_01: But we're going to talk about how these things aren't really true.
• 5:39
  SPEAKER_01: So maybe bring them in on this conversation.
• 5:41
  SPEAKER_01: Okay, so let's jump into some various um scary things that very intelligent people have said.
• 5:50
  SPEAKER_01: And this is one of those things that too often people think that tech CEOs are somehow much smarter than the average Joe.
• 5:59
  SPEAKER_01: And that they are somehow gifted with this ability that can see into the future, can see how things are going to work, and that they know so much more than the average developer.
• 6:09
  SPEAKER_01: So therefore, when they say something, it must be the truth.
• 6:12
  SPEAKER_01: And the fact is, it's almost exactly the opposite.
• 6:16
  SPEAKER_01: I'm not saying they're dumb.
• 6:18
  SPEAKER_01: What I'm saying is they're not talking to you.
• 6:21
  SPEAKER_01: They're not talking about reality.
• 6:23
  SPEAKER_01: What they're talking about is trying to boost their stock price.
• 6:27
  SPEAKER_01: That's their whole goal.
• 6:28
  SPEAKER_01: So when they say something, it's not to give an actual prediction of the future, it's to give an actual boost to their stock.
• 6:37
  SPEAKER_01: That's basically it.
• 6:40
  SPEAKER_01: They're not smarter than the average Joe.
• 6:42
  SPEAKER_01: In fact, in a lot of times, what they're saying isn't connected with reality.
• 6:47
  SPEAKER_01: So let's talk about this.
• 6:49
  SPEAKER_01: Let's talk first with the Microsoft AI CEO.
• 6:52
  SPEAKER_01: Now, this is not Satya Nadella, this is just the head of AI at Microsoft.
• 6:58
  SPEAKER_01: So he said that AI can replace every white-collar job in 18 months.
• 7:04
  SPEAKER_01: Now, this is um as of, well, this is five hours ago, but this is in January.
• 7:09
  SPEAKER_01: Um this is his prediction.
• 7:14
  SPEAKER_01: 18 months from now, no white-collar job.
• 7:17
  SPEAKER_01: That means no lawyers, that means no accountants, that means no software developers.
• 7:22
  SPEAKER_01: AI is gonna do it all.
• 7:24
  SPEAKER_01: Now, that sounds terrifying.
• 7:27
  SPEAKER_01: Because if that were true, first of all, who's paying for the AI at this point?
• 7:31
  SPEAKER_01: Just the CEOs are gonna be buying, you know, a CEO's gonna be one person, they're gonna own a whole company full of AIs, maybe?
• 7:38
  SPEAKER_01: And then who's gonna buy their products?
• 7:40
  SPEAKER_01: Because we're all gonna be broke because we have no jobs, or we're gonna go back to digging ditches or something.
• 7:45
  SPEAKER_01: I'm not sure what.
• 7:46
  SPEAKER_01: Um, that sounds terrifying.
• 7:48
  SPEAKER_01: And you know what?
• 7:50
  SPEAKER_01: It would be terrifying if he was actually trying to say something real here.
• 7:54
  SPEAKER_01: He's not.
• 7:55
  SPEAKER_01: What he's trying to do is boost his stock.
• 7:59
  SPEAKER_01: Okay.
• 8:00
  SPEAKER_01: So the end is near, right?
• 8:02
  SPEAKER_01: Um and he says 12 to 18 months actually.
• 8:05
  SPEAKER_01: So he does this art this interview with Financial Times where he talks about you know super intelligence and all the rest.
• 8:12
  SPEAKER_01: 18 months.
• 8:13
  SPEAKER_01: Now, let's look at this same exact person a couple of years ago.
• 8:19
  SPEAKER_01: So he is predicting 18 months, 12 to 18 months actually, to replace every white-collar job, which is impossible.
• 8:27
  SPEAKER_01: But especially since he is working for the company that built Copilot.
• 8:31
  SPEAKER_01: And we all know that Copilot is nowhere near ready for sentience.
• 8:35
  SPEAKER_01: But let's look at what he said a couple of years ago.
• 8:38
  SPEAKER_01: Because if what he said a couple of years ago came true, this is more, uh, I'm more willing to believe what he might say.
• 8:46
  SPEAKER_01: Okay.
• 8:47
  SPEAKER_01: But if he is basically telling lies uh two years ago, then he's probably telling lies again.
• 8:54
  SPEAKER_01: Okay?
• 8:54
  SPEAKER_01: Let's look at the pattern here.
• 8:55
  SPEAKER_01: So it's the same person two years ago.
• 8:57
  SPEAKER_01: This is a Twitter.
• 8:59
  SPEAKER_01: LLM hallucinations we largely eliminated by 2025.
• 9:03
  SPEAKER_01: That was last year.
• 9:04
  SPEAKER_01: Largely eliminated.
• 9:06
  SPEAKER_01: That's a huge deal.
• 9:07
  SPEAKER_01: The implications are far more profound than the threat of the models getting things a bit wrong today.
• 9:12
  SPEAKER_01: What he's saying is, hey, yeah, I know I get things wrong today, but you know what?
• 9:15
  SPEAKER_01: In two years, there won't be hallucinations, or practically no hallucinations.
• 9:20
  SPEAKER_01: So that was two years ago.
• 9:24
  SPEAKER_01: This was uh June of 2023, so almost three years ago, two and a half years ago.
• 9:29
  SPEAKER_01: He said this.
• 9:30
  SPEAKER_01: Now, is that true?
• 9:31
  SPEAKER_01: And the first thing I've heard people say is, well, it feels true.
• 9:34
  SPEAKER_01: There's less hallucinations now than there were.
• 9:37
  SPEAKER_01: And actually, that's not true.
• 9:40
  SPEAKER_01: They're more insidious about their hallucinations.
• 9:43
  SPEAKER_01: In fact, now OpenAI admits that hallucinations are not, are mathematically inevitable, not just an engineering flaw.
• 9:53
  SPEAKER_01: So this is this came out in September of 2025.
• 9:56
  SPEAKER_01: So the year that we are supposed to have no hallucinations or almost no hallucinations, OpenAI admitted that they're mathematically inevitable.
• 10:06
  SPEAKER_01: They cannot remove hallucinations.
• 10:08
  SPEAKER_01: In fact, there is a whole lot of stuff I didn't go into in this article.
• 10:14
  SPEAKER_01: Um, you can read more about it if you want.
• 10:16
  SPEAKER_01: Um, but this is from Computer World.
• 10:19
  SPEAKER_01: But hallucinations now are really bad.
• 10:22
  SPEAKER_01: In fact, I saw one article that said that for all the major models right now, if you say, Are you sure?
• 10:28
  SPEAKER_01: 60% of the time, they will change their answers.
• 10:32
  SPEAKER_01: That's a hallucination.
• 10:34
  SPEAKER_01: That's saying, oh, yeah, what I say before was wrong.
• 10:38
  SPEAKER_01: This is the right way.
• 10:38
  SPEAKER_01: Well, which is it?
• 10:40
  SPEAKER_01: One of you is wrong.
• 10:41
  SPEAKER_01: The current you or the previous you from five minutes ago.
• 10:45
  SPEAKER_01: That's a problem.
• 10:47
  SPEAKER_01: So when he said that two years ago, two and a half years ago, he said that all hallucinations are practically all hallucinations, largely eliminated by 2025.
• 10:59
  SPEAKER_01: Totally not true.
• 11:00
  SPEAKER_01: Not even close.
• 11:01
  SPEAKER_01: And in fact, mathematically inevitable.
• 11:04
  SPEAKER_01: It's impossible to remove hallucinations according to OpenAI.
• 11:08
  SPEAKER_01: And that's what we've seen so far.
• 11:10
  SPEAKER_01: So when he says now, 18 months from now, we're gonna have no white-collar jobs not done by AI, that's about the same category.
• 11:21
  SPEAKER_01: It's not gonna happen.
• 11:22
  SPEAKER_01: But let's see this pattern continue with other CEOs and other high-level people.
• 11:28
  SPEAKER_01: Let's go to anthropic next.
• 11:30
  SPEAKER_01: So, Anthropic CEO, this is uh January of 2026, so just a month ago.
• 11:37
  SPEAKER_01: Anthropic CEO predicts AI models will replace software engineers in six to 12 months.
• 11:43
  SPEAKER_01: So he says, I don't write code anymore, which that's that's fun because why is a CEO writing code to begin with, right?
• 11:50
  SPEAKER_01: Like, yes, maybe you were originally a coder and you start a company, but at some point as a CEO, you have to stop writing code.
• 11:57
  SPEAKER_01: Like, not we if you're a large company or large enough like Anthropic is, you have better things to do.
• 12:05
  SPEAKER_01: But he says within six to 12 months, there'll be no software engineers.
• 12:11
  SPEAKER_01: We don't need them.
• 12:12
  SPEAKER_01: Okay, so that sounds scary.
• 12:15
  SPEAKER_01: And you know what?
• 12:16
  SPEAKER_01: That would mean that by June to December of 2026, somewhere in that time range, we all go away.
• 12:23
  SPEAKER_01: We have to go look for jobs somewhere else.
• 12:26
  SPEAKER_01: But let's take a peek back at a previous prediction the same person had.
• 12:31
  SPEAKER_01: So Anthropic CEO, same CEO, says that in three to six months, AI will be writing 90% of the code software developers are in charge of.
• 12:41
  SPEAKER_01: And further, he says that AI could be writing essentially all the code in 12 months.
• 12:48
  SPEAKER_01: Now, when was this published?
• 12:51
  SPEAKER_01: March of 2025.
• 12:54
  SPEAKER_01: That means that 90% of the code, at his high estimate, um, 90% of the code written by AI should have been done by what September timeframe?
• 13:05
  SPEAKER_01: And we're closing in on a year.
• 13:08
  SPEAKER_01: At the time of this recording, we're about two weeks away from a year from when he said a year ago from when he said this.
• 13:15
  SPEAKER_01: So at this point, we should have almost no software development jobs, or at least no developers writing code, it should be all AI.
• 13:24
  SPEAKER_01: Now you may think, well, it does feel like everyone's writing AI code.
• 13:28
  SPEAKER_01: And yes, it does feel that way.
• 13:30
  SPEAKER_01: But there is a lot of software development being written.
• 13:35
  SPEAKER_01: And no, AI is not replacing software developers.
• 13:39
  SPEAKER_01: We'll see more about that.
• 13:41
  SPEAKER_01: But basically, even though people are using AI to write a lot of code, what they find is they have to do more work on it than if they wrote it themselves.
• 13:52
  SPEAKER_01: So, no, we have not replaced software developers with AI.
• 13:56
  SPEAKER_01: People keep trying.
• 13:57
  SPEAKER_01: There's still headlines like that first one where you know it replaces all the work of a year and one hour, that kind of stuff.
• 14:05
  SPEAKER_01: But it's not happened.
• 14:07
  SPEAKER_01: So this didn't happen.
• 14:10
  SPEAKER_01: He said in 12 months, basically all software written by AI.
• 14:14
  SPEAKER_01: So now, almost 12 months later, about you know, 10 months later, he says, well, it's actually six to 12 months from now.
• 14:24
  SPEAKER_01: So he didn't kind of address the fact that he's already said this.
• 14:29
  SPEAKER_01: He just said it again.
• 14:30
  SPEAKER_01: And he gave a new deadline.
• 14:32
  SPEAKER_01: And it's always a new deadline, it's always six to twelve months, it's always 12 months to 18 months, it's always this just around the corner thing, but it never actually arrives.
• 14:43
  SPEAKER_01: But let's keep going and see this pattern continue.
• 14:46
  SPEAKER_01: Let's go to Salesforce.
• 14:48
  SPEAKER_01: Salesforce CEO confirms 4,000 layoffs were because I needed less heads with AI.
• 14:55
  SPEAKER_01: So Salesforce laid off a bunch of people, 4,000 people, and they said, well, we don't need them because we have AI.
• 15:04
  SPEAKER_01: This was in September of 2025, not that long ago.
• 15:10
  SPEAKER_01: Now, let's fast forward to the same year, December.
• 15:14
  SPEAKER_01: So from September of 2025 to December of 2025.
• 15:18
  SPEAKER_01: After laying off 4,000 employees and automating with AI agents, Salesforce executives admit we were more competent or confident than prepared, and competent probably too.
• 15:30
  SPEAKER_01: Um, so they were not prepared for the fact that it didn't work.
• 15:36
  SPEAKER_01: They thought it was going to work.
• 15:37
  SPEAKER_01: They made a big splashy headline, but they didn't even last half a year before he said, oops, that was a mistake.
• 15:46
  SPEAKER_01: And they go into more about the fact that yes, there's still AI and all the rest, but they realize they have a problem because AI doesn't do all the things they thought it was going to do.
• 15:56
  SPEAKER_01: So they taught where automation fell short and the tension that it caused and a strategic reset.
• 16:04
  SPEAKER_01: So what they did was they made a big splashy headline.
• 16:08
  SPEAKER_01: And you know what?
• 16:09
  SPEAKER_01: That changes stock prices.
• 16:10
  SPEAKER_01: And that makes it look like they're this forward-thinking company who is cutting down their costs they don't need to and replacing people with AI because that's gonna make more money for the shareholders.
• 16:21
  SPEAKER_01: And so the shareholders are happy.
• 16:23
  SPEAKER_01: However, what it doesn't do is what they think it's gonna do.
• 16:28
  SPEAKER_01: Now, in this case, I don't think they were hiding.
• 16:32
  SPEAKER_01: You know, some companies will lay people off and say, well, it's because we don't need them because of AI.
• 16:37
  SPEAKER_01: What they're really doing is just hiding or giving a different name to a layoff.
• 16:41
  SPEAKER_01: Like maybe they were hurting financially and they use AI as a cover so they can actually boost stock while fixing a problem in their books.
• 16:50
  SPEAKER_01: I don't think this is the case.
• 16:51
  SPEAKER_01: I think this is the case where they genuinely thought that AI was the solution to their overstaffing.
• 16:59
  SPEAKER_01: And so you say, you know what, let's just make AI do the work and we can, you know, lay off these people.
• 17:05
  SPEAKER_01: These aren't software developers they laid off.
• 17:06
  SPEAKER_01: These are just regular people.
• 17:08
  SPEAKER_01: And it still didn't work.
• 17:10
  SPEAKER_01: So that's self-Salesforce.
• 17:12
  SPEAKER_01: Again, making a prediction and then and even backing it up with a significant change in your company and then falling flat in your face because no, you're wrong.
• 17:23
  SPEAKER_01: But let's look at Meta.
• 17:25
  SPEAKER_01: Because again, we're talking about CEOs and how you know they're not really as smart as we think they are.
• 17:30
  SPEAKER_01: And the very fact that I have this tab called Meta should give you a clue of where we're going here.
• 17:37
  SPEAKER_01: This is still on Meta's website.
• 17:40
  SPEAKER_01: Now, if you don't know, Meta is Facebook, but they renamed to Meta.
• 17:44
  SPEAKER_01: Why?
• 17:45
  SPEAKER_01: Because the metaverse was going to be where we all lived.
• 17:49
  SPEAKER_01: Now, this is from the 2023 Future of Work Report.
• 17:55
  SPEAKER_01: This is less than three years old.
• 17:59
  SPEAKER_01: This is the future of work.
• 18:00
  SPEAKER_01: Let's come down here to the four pillars here.
• 18:03
  SPEAKER_01: Number one, inclusion is vital for remote work.
• 18:06
  SPEAKER_01: Okay, so a digital experience along the way, fine.
• 18:10
  SPEAKER_01: Number two, more businesses are investing in VR.
• 18:14
  SPEAKER_01: Really?
• 18:15
  SPEAKER_01: Is your business investing in VR?
• 18:17
  SPEAKER_01: According to this, 60% of businesses want to see VR headsets introduced to their workforce by 2024.
• 18:25
  SPEAKER_01: Again, 12 months from this prediction.
• 18:29
  SPEAKER_01: Now, do you see your employer investing in VR?
• 18:34
  SPEAKER_01: Is your employer, you know, lining up headsets and saying, hey, here's the headset instead of your monitor station?
• 18:40
  SPEAKER_01: No, I don't think that's happening pretty much anywhere.
• 18:44
  SPEAKER_01: Now, number three, interoperability for better collaboration.
• 18:49
  SPEAKER_01: Businesses have allocated an average of$24 million for VR or AR investments.
• 18:58
  SPEAKER_01: Around 15% of all tech spent has been on VR and AR for companies.
• 19:03
  SPEAKER_01: I'm not sure what companies they're talking about, because this didn't seem to happen at all.
• 19:09
  SPEAKER_01: There were a few companies that tried things, did trials, and all the rest.
• 19:12
  SPEAKER_01: This didn't happen.
• 19:14
  SPEAKER_01: Okay.
• 19:15
  SPEAKER_01: And this is the one I really want you to focus on.
• 19:18
  SPEAKER_01: Number four, metaverse skeptics will miss out.
• 19:21
  SPEAKER_01: It's the fear of missing out.
• 19:23
  SPEAKER_01: If you don't get on board now, you're gonna miss this massive trend, which sounds eerily familiar to if you don't get on board with AI now, you're gonna miss out.
• 19:33
  SPEAKER_01: If you're not up to date in the latest Claude or Copilot or Gemini or whatever else, you're gonna somehow miss out as a developer or as an employee of a regular company.
• 19:45
  SPEAKER_01: That's not true.
• 19:46
  SPEAKER_01: Okay.
• 19:47
  SPEAKER_01: But how can firms ensure that their employees be open to exploring the metaverse?
• 19:53
  SPEAKER_01: And what if they don't act in time?
• 19:55
  SPEAKER_01: Some 70% of workers expect the metaverse to allow for more global hiring.
• 20:01
  SPEAKER_01: And it's kind of funny, they tacked out of the end, hey, we'll have more global hiring if you have VR.
• 20:05
  SPEAKER_01: But this whole concept is the metaverse is going to change how we do things.
• 20:13
  SPEAKER_01: And again, Facebook changed the name of the entire company to Meta to better align with their vision of the metaverse is where things are at.
• 20:27
  SPEAKER_01: This has only been three years.
• 20:30
  SPEAKER_01: And even now we're looking back and going, really?
• 20:33
  SPEAKER_01: They did that?
• 20:34
  SPEAKER_01: Like I forgot all about that.
• 20:36
  SPEAKER_01: Because it was a hype thing.
• 20:39
  SPEAKER_01: And there was a lot of money invested in it.
• 20:42
  SPEAKER_01: And yet it didn't really go anywhere.
• 20:44
  SPEAKER_01: Let's look at the 12 top metaverse predictions.
• 20:48
  SPEAKER_01: This is from Tech Target.
• 20:49
  SPEAKER_01: This was back in 2020, the end of 2022.
• 20:54
  SPEAKER_01: So prediction one, the metaverse will be big, but just how big?
• 20:59
  SPEAKER_01: Okay.
• 21:00
  SPEAKER_01: So they're already saying, you know what, we we know it's gonna be big, but just how big is this thing gonna be?
• 21:06
  SPEAKER_01: It's gonna generate, and I love this, uh, McKinsey Company, which McKinsey and Company is a consulting firm and always seems to be around the worst companies, um, predicted that metaverse economy could, in aggregate, generate up to five trillion dollars impact by 2030, equivalent to the size of the third largest economy today, Japan.
• 21:30
  SPEAKER_01: That didn't happen, okay?
• 21:32
  SPEAKER_01: But the predictions were five trillion by 2030.
• 21:37
  SPEAKER_01: Right now, the AI companies are saying two trillion dollars for AI by 2030.
• 21:45
  SPEAKER_01: Sounds kind of familiar.
• 21:47
  SPEAKER_01: So that's the first one.
• 21:48
  SPEAKER_01: Number two, there'll be multiple metaverses.
• 21:52
  SPEAKER_01: Okay, first of all, it's gonna be big.
• 21:55
  SPEAKER_01: But not only big, there's gonna be multiple of them.
• 21:58
  SPEAKER_01: Again, it didn't happen.
• 22:00
  SPEAKER_01: Number three, metaverse will transform all business processes.
• 22:03
  SPEAKER_01: Sound familiar?
• 22:04
  SPEAKER_01: Like AI is gonna transform everything you do by because we're gonna have copilot and everything.
• 22:09
  SPEAKER_01: And everything you do is gonna have an AI.
• 22:12
  SPEAKER_01: You know what?
• 22:13
  SPEAKER_01: It's not nearly as useful as people thought initially it would be.
• 22:16
  SPEAKER_01: And we're finding that out.
• 22:17
  SPEAKER_01: More and more people hate AI.
• 22:20
  SPEAKER_01: And the biggest feature about copilot and all the other things that people are looking for is how do I remove it?
• 22:28
  SPEAKER_01: So the prediction that it will transform business processes is eerily familiar to the metaverse.
• 22:36
  SPEAKER_01: Number four, new security and privacy issues will emerge.
• 22:39
  SPEAKER_01: That's true about AI as well.
• 22:40
  SPEAKER_01: Number five, the metaverse will expand beyond virtual showrooms.
• 22:45
  SPEAKER_01: It didn't.
• 22:46
  SPEAKER_01: Number six, digital twins will mirror the entire enterprise.
• 22:49
  SPEAKER_01: The idea that you're gonna have a virtual you and it's gonna be everywhere.
• 22:54
  SPEAKER_01: And then down here, the metaverse's role in a future business and all these percentage of places that it impact.
• 23:00
  SPEAKER_01: Number seven, customer service and sales roles will transform and automate.
• 23:06
  SPEAKER_01: The idea that this is going to basically remove the need for salespeople didn't happen.
• 23:12
  SPEAKER_01: Digital first will evolve into metaverse.
• 23:17
  SPEAKER_01: Didn't happen.
• 23:18
  SPEAKER_01: Enterprises will bake reality into the metaverse.
• 23:21
  SPEAKER_01: Didn't happen.
• 23:22
  SPEAKER_01: The metaverse will consolidate technologies.
• 23:25
  SPEAKER_01: Sounds like AI as well.
• 23:27
  SPEAKER_01: Gamification will galvanize calm technologies.
• 23:30
  SPEAKER_01: I'm not even sure how they're thinking about that, but that's fine.
• 23:34
  SPEAKER_01: And robotics innovation will accelerate.
• 23:36
  SPEAKER_01: Again, what AI is promising as well.
• 23:39
  SPEAKER_01: In fact, you know, Tesla is shutting down certain plants to change them over to be making robots instead.
• 23:46
  SPEAKER_01: I wouldn't invest in those if that were you.
• 23:49
  SPEAKER_01: So these are the predictions for the metaverse.
• 23:53
  SPEAKER_01: Again, these predictions are less than four years old, but uh just over three years old.
• 23:58
  SPEAKER_01: And yet totally wrong.
• 24:01
  SPEAKER_01: But they were said with great confidence.
• 24:04
  SPEAKER_01: Again, if we look at prediction number one, it's not even that it's going to be big, but like they're saying this is a given.
• 24:12
  SPEAKER_01: It's gonna be a given that's gonna be big.
• 24:15
  SPEAKER_01: But just how big?
• 24:16
  SPEAKER_01: That that arrogance of we know for sure it's gonna get this gonna take off.
• 24:21
  SPEAKER_01: And reality is here's reality.
• 24:24
  SPEAKER_01: Meta has discontinued its metaverse for work too.
• 24:28
  SPEAKER_01: They shut off the metaverse for people, and they shut it off for work as well, because it didn't go anywhere.
• 24:35
  SPEAKER_01: It flopped.
• 24:36
  SPEAKER_01: It had a huge beginning, it had a big splash, and big tech companies were trying to push it, and now it's gone.
• 24:44
  SPEAKER_01: And now the next shiny thing is on the horizon.
• 24:48
  SPEAKER_01: So that's meta.
• 24:50
  SPEAKER_01: And again, the idea that these big predictions and how we're gonna revolutionize things isn't really true.
• 24:59
  SPEAKER_01: And it really was to kind of uh make something happen by trying to build momentum.
• 25:06
  SPEAKER_01: Okay, let's look at this next category.
• 25:08
  SPEAKER_01: And this next category is actually just one Wikipedia list because I didn't want to go diving into all the predictions.
• 25:14
  SPEAKER_01: But one of the people, and I don't want to get into anything too deep on this, but one of the people that is looked to as a super smart genius for technology is Elon Musk.
• 25:26
  SPEAKER_01: And the idea that whatever he says is going to happen, like changing over the uh the plants that were making certain Teslas, and now they're making uh robots that still don't work on their own.
• 25:40
  SPEAKER_01: But they're going to soon.
• 25:43
  SPEAKER_01: So let's look at just a few of the predictions that he has made.
• 25:50
  SPEAKER_01: So this is a Wikipedia article that is the list of predictions for autonomous Tesla vehicles by Elon Musk.
• 25:57
  SPEAKER_01: And again, what's Elon's job?
• 26:00
  SPEAKER_01: It isn't to write code, it isn't to even tell you the truth.
• 26:05
  SPEAKER_01: His job is to boost the stock.
• 26:09
  SPEAKER_01: And he has done a spectacular job at that.
• 26:12
  SPEAKER_01: In fact, Tesla is worth more than uh Toyota as far as uh market cap.
• 26:19
  SPEAKER_01: And yet it has less assets.
• 26:21
  SPEAKER_01: And why is that?
• 26:22
  SPEAKER_01: Because there is a lot of hype around what people think can be done.
• 26:27
  SPEAKER_01: And to be fair, and I'm I want to be very clear here, I'm not bashing the fact that I really like the idea of full self-drive.
• 26:35
  SPEAKER_01: I like the idea that a car can drive on its own, but let's talk reality of that.
• 26:41
  SPEAKER_01: Because in 2013, we're talking, what is that now, 13 years ago, uh, 12 and a half years ago?
• 26:50
  SPEAKER_01: He said we should be able to do 90% of miles driven autonomously within three years.
• 26:57
  SPEAKER_01: That would mean that in 2016 we'd have 90% of the miles driven would be done autonomously.
• 27:05
  SPEAKER_01: Didn't happen, not even close.
• 27:07
  SPEAKER_01: We can go down this list of prediction after prediction after and notice the was it met?
• 27:13
  SPEAKER_01: No um kind of.
• 27:15
  SPEAKER_01: Um, but notice this was a prediction from 2014 that was supposed to be for 2015.
• 27:21
  SPEAKER_01: Okay.
• 27:22
  SPEAKER_01: Um all of this, no, this prediction from August to December was correct, or mostly.
• 27:32
  SPEAKER_01: Um no, there's one yes here, and that was robot in June of 2025.
• 27:38
  SPEAKER_01: He said that in June of 2025, robotaxis would be launched.
• 27:43
  SPEAKER_01: He was close, he was six days off.
• 27:45
  SPEAKER_01: Um, so yes.
• 27:47
  SPEAKER_01: And no or to be determined.
• 27:50
  SPEAKER_01: So this is the track record of a person that when he speaks, people say that must be what's going to happen.
• 27:58
  SPEAKER_01: So when you see things like him saying, well, AI is going to be able to do this or do that, remember this track record of no, those things didn't happen.
• 28:10
  SPEAKER_01: And again, the full self-drive isn't actually full self-drive yet, by the definition of what he wanted.
• 28:17
  SPEAKER_01: It's close, it's getting better, but nowhere near the predictions of 2013 and beyond.
• 28:26
  SPEAKER_01: So when you look at predictions, don't just take them with a grain of salt.
• 28:32
  SPEAKER_01: Don't believe them.
• 28:34
  SPEAKER_01: Look at what's actually happening.
• 28:36
  SPEAKER_01: So let's look at a few results, and we're gonna talk more about what we can do.
• 28:41
  SPEAKER_01: This so far has just been they don't tell the truth, or the truth they're telling is just here's how to boost stock price, not actually what's going to happen.
• 28:52
  SPEAKER_01: So let's look at results.
• 28:54
  SPEAKER_01: Um, this is from uh from Twitter.
• 28:56
  SPEAKER_01: Software reliability is down across the industry.
• 29:00
  SPEAKER_01: So reliability, not amount of lines of code written, reliability.
• 29:06
  SPEAKER_01: Failure rate is up, and batch size is up.
• 29:10
  SPEAKER_01: So the amount of code written across the industry is going up.
• 29:16
  SPEAKER_01: The batch size is going up.
• 29:18
  SPEAKER_01: More big chunks of code are being released, but yet the reliability is down and the failure rate is up.
• 29:24
  SPEAKER_01: So AI is generating larger changes, and we know from research, from just larger changes in general, not just AI, that larger batch sizes tend to result in exactly these problems of bad reliability and high failure rate.
• 29:41
  SPEAKER_01: So this is what's happening.
• 29:43
  SPEAKER_01: Yes, AI is being used to write code, but it's unreliable code that is more prone to failure.
• 29:52
  SPEAKER_01: In fact, I didn't include this one, but it has gotten so bad on GitHub.
• 29:57
  SPEAKER_01: GitHub itself turned off its uptime percentage.
• 30:03
  SPEAKER_01: We used to talk 10 years ago or so, and it's kind of gone off favor to talk about too much.
• 30:08
  SPEAKER_01: We talked about uptime percentages, the idea of how many nines of availability do you have, which is 99%, 99.9%, 99.99%, and so on.
• 30:20
  SPEAKER_01: And the goal has always been five nines for most applications.
• 30:24
  SPEAKER_01: Now it's incredibly hard to do.
• 30:27
  SPEAKER_01: Five nines, I believe, is like eight hours of downtime a year to have five nines of availability.
• 30:33
  SPEAKER_01: Right now, on a I believe it's a 90-day uh rolling basis.
• 30:44
  SPEAKER_01: 92, which means they have one nine of availability.
• 30:51
  SPEAKER_01: That's a major problem.
• 30:52
  SPEAKER_01: Now it's not every service that's a cumulative of their services, but it was so bad that they took down their status page and changed it so it doesn't show you the percentages anymore because it's too ugly and it's not getting better.
• 31:09
  SPEAKER_01: Why?
• 31:10
  SPEAKER_01: Because AI is writing more code.
• 31:12
  SPEAKER_01: We'll talk why this is a great thing going forward, but just know AI writes bad code.
• 31:19
  SPEAKER_01: CodeRabbit did the state of AI versus human code generation report.
• 31:24
  SPEAKER_01: And I would encourage you to read this report.
• 31:26
  SPEAKER_01: This report is fascinating, it's got a lot of good stuff in here.
• 31:29
  SPEAKER_01: But here's the overall conclusion they came to.
• 31:33
  SPEAKER_01: We found AI written code produces 1.7 times more issues.
• 31:40
  SPEAKER_01: So if you use AI to write code, it's gonna generate 1.7 times or 70% more bugs than if you wrote yourself.
• 31:52
  SPEAKER_01: That's not a good thing.
• 31:54
  SPEAKER_01: That's not a software developer's gonna go away anytime soon thing.
• 31:59
  SPEAKER_01: And it's not like you can say, just make no errors.
• 32:02
  SPEAKER_01: That doesn't work.
• 32:03
  SPEAKER_01: Um, and they break down what kind of errors.
• 32:06
  SPEAKER_01: This is where it gets really important because the types of errors are actually really scary.
• 32:13
  SPEAKER_01: Because security issues are 1.5 times higher, and two times more improper passwords were handled when using AI.
• 32:26
  SPEAKER_01: That's a big deal.
• 32:27
  SPEAKER_01: Security is a big deal.
• 32:29
  SPEAKER_01: Um, I want to point out number seven, um performance regressions, which means things slow down.
• 32:37
  SPEAKER_01: It says that they're rare.
• 32:38
  SPEAKER_01: So be careful these numbers.
• 32:40
  SPEAKER_01: Don't just take this number and say, oh, you know, the number says that that AIs create eight times more issues with performance.
• 32:50
  SPEAKER_01: But the reality is there weren't very many, so eight times more is a hard number to actually, it's not statistically relevant, probably.
• 32:58
  SPEAKER_01: Um, but in general, logic errors, um, escalation errors, readability concerns, which you may say, why does readability matter?
• 33:09
  SPEAKER_01: So readability is important because of the fact that AI creates 1.7 times more errors than humans do.
• 33:18
  SPEAKER_01: And humans create a whole bunch of errors.
• 33:20
  SPEAKER_01: So by that logic, you need to be able to read the code to understand where those errors are.
• 33:26
  SPEAKER_01: And this is the line that really stands out to me.
• 33:28
  SPEAKER_01: AI code looks right at a glance.
• 33:32
  SPEAKER_01: It looks right at a glance, but often violates local idioms or structure.
• 33:37
  SPEAKER_01: So it's saying is there's a real problem here because it looks right.
• 33:42
  SPEAKER_01: So um the this report is great.
• 33:45
  SPEAKER_01: It goes into more details, it really breaks down these different things.
• 33:48
  SPEAKER_01: Now, yes, you can say, well, but the latest model, and that's always the excuse, right?
• 33:55
  SPEAKER_01: The excuse is, well, that's true, but you know, the latest model doesn't do that.
• 33:59
  SPEAKER_01: And this was from, I don't remember what the this report is from.
• 34:03
  SPEAKER_01: Um, I'm not sure if it says it's in 2025.
• 34:06
  SPEAKER_01: Um, so you know what?
• 34:08
  SPEAKER_01: Um it's 2026 now, and there's there's new models and there's new ways of doing things, and they're so much better.
• 34:13
  SPEAKER_01: And you hear that all the time.
• 34:15
  SPEAKER_01: We've been hearing that for years.
• 34:18
  SPEAKER_01: It's always been, but the next model.
• 34:22
  SPEAKER_01: And the reality is these things aren't getting better.
• 34:25
  SPEAKER_01: We keep doing these reports.
• 34:27
  SPEAKER_01: I saw these reports in 2023, I saw them in 2024, I saw them now in 2025.
• 34:33
  SPEAKER_01: We're gonna see them again in 2026 because it doesn't actually change.
• 34:39
  SPEAKER_01: So AI writes bad code.
• 34:44
  SPEAKER_01: This one is an interesting report because there's there is a bit of, you don't want to go too deep into reading into this report because there's not enough data in here to be statistically relevant enough.
• 34:58
  SPEAKER_01: Um but what they this this um study found is that people think that AI will make them faster, but it actually doesn't, makes them slower.
• 35:11
  SPEAKER_01: So that's the basic conclusion.
• 35:12
  SPEAKER_01: You can come down here to core results down here and see um that when AI was not allowed uh versus what they forecasted.
• 35:21
  SPEAKER_01: So you say, hey, I think this thing's gonna take me an hour.
• 35:25
  SPEAKER_01: And then you actually do it, but you don't use AI, it took you go a little bit faster than an hour, probably.
• 35:31
  SPEAKER_01: That's what this is saying.
• 35:33
  SPEAKER_01: And then if you said, hey, it's the same thing, I might take it.
• 35:36
  SPEAKER_01: This probably took an hour using AI, well, it took an hour and a half or something.
• 35:41
  SPEAKER_01: Like this is what it's saying is that you expect it to take this amount of time, but it takes a bit longer.
• 35:46
  SPEAKER_01: Um now again, the statistics of this are not deep enough to say for certain this applies to everything.
• 35:54
  SPEAKER_01: And yes, it's really hard to measure these things when it comes to software development because how do you measure that it should take this long and it took this long, and how it would what we would have done if it took an AI to do it, and so on.
• 36:06
  SPEAKER_01: So take a grain of salt.
• 36:08
  SPEAKER_01: Don't just look at a number and go, this supports my argument, therefore it must be right.
• 36:13
  SPEAKER_01: Don't do that.
• 36:14
  SPEAKER_01: This just gives us an indication that, yeah, maybe we're um we're looking at AI not doing as much as we think it does, which pairs well with the concept of AI makes more mistakes than we do, and that AI has a bad failure rate and a low reliability in the software it creates.
• 36:40
  SPEAKER_01: Okay?
• 36:41
  SPEAKER_01: Now let's look at cursor CEO.
• 36:42
  SPEAKER_01: You probably may not have heard of cursor for a while.
• 36:45
  SPEAKER_01: Um, cursor has been on off the radar.
• 36:48
  SPEAKER_01: It was really big initially, and that's kind of backed off a bit.
• 36:51
  SPEAKER_01: But the cursor CEO, which again, remember, don't just trust CEOs because they're a CEO.
• 36:58
  SPEAKER_01: So I want to point this out again, grand assault.
• 37:01
  SPEAKER_01: But the cursor CEO warns that vibe coding could lead to unstable software foundations.
• 37:08
  SPEAKER_01: So you can read more about this.
• 37:10
  SPEAKER_01: Um, but I agree with that because of the fact that we see that it causes 1.7 times more issues.
• 37:20
  SPEAKER_01: The fact that batch size is up, failure rate is up, and reliability is down.
• 37:26
  SPEAKER_01: So this kind of that statement lines up well with the facts that we're seeing.
• 37:33
  SPEAKER_01: So therefore, it seems to be in line, it seems to be correct.
• 37:38
  SPEAKER_01: Now, let's talk about the hope side of this.
• 37:40
  SPEAKER_01: Because all this so far we've seen that CEOs don't tell a truth or they're delusional and they're looking just to raise stock prices.
• 37:48
  SPEAKER_01: And maybe they're just optimistic.
• 37:50
  SPEAKER_01: Maybe it's a better way of saying it said delusional.
• 37:51
  SPEAKER_01: They're optimistic.
• 37:53
  SPEAKER_01: Um, they're wrong.
• 37:54
  SPEAKER_01: But even though they say these things we know are not true, because we've seen over and over again they haven't been true.
• 38:01
  SPEAKER_01: They have a track record of being incorrect.
• 38:03
  SPEAKER_01: And we see now that AI is writing bad code and it's writing non-performant code, and it's causing problems with the code that it writes, and it makes it software developers slower and creates less reliability.
• 38:18
  SPEAKER_01: But where does that leave us?
• 38:20
  SPEAKER_01: Well, people are starting to catch on.
• 38:23
  SPEAKER_01: Right now, there's a lot of uh people saying, hey, I want to use AI as an excuse to lay people off, or I want to try to cut my bottom line, or maybe even I believe the AI hype, and so I'm gonna let developers go.
• 38:40
  SPEAKER_01: And that has caused some problems in the industry, but we're starting to see a resurgence of people going, wait, that's dumb.
• 38:48
  SPEAKER_01: That's incorrect.
• 38:49
  SPEAKER_01: That's the wrong way to go.
• 38:51
  SPEAKER_01: And in fact, this one right here, I love this.
• 38:53
  SPEAKER_01: IBM is tripling the number of Gen Z entry-level jobs after finding the limits of AI adoption.
• 39:03
  SPEAKER_01: Too often we hear that junior developers are going to go away because AI replaces them.
• 39:10
  SPEAKER_01: And the thought is that AI is like a junior developer.
• 39:14
  SPEAKER_01: The reality is it's nothing like a junior developer because a junior developer can grow and change and learn.
• 39:21
  SPEAKER_01: AIs don't.
• 39:23
  SPEAKER_01: So that's a problem.
• 39:25
  SPEAKER_01: But companies are trying to believe that.
• 39:27
  SPEAKER_01: But IBM says, wait, wait, nope.
• 39:29
  SPEAKER_01: We believe that, and now we don't believe that anymore.
• 39:32
  SPEAKER_01: And in fact, we're going to triple the number of entry-level jobs for software developers because it's that important.
• 39:41
  SPEAKER_01: So that's one thing.
• 39:43
  SPEAKER_01: But let's look at the economy as a whole, because too often I hear people say, you know what, it's just too dangerous to get into software development.
• 39:52
  SPEAKER_01: You should be a plumber.
• 39:53
  SPEAKER_01: You should be an electrician.
• 39:54
  SPEAKER_01: Get in the trades.
• 39:56
  SPEAKER_01: Um, and I usually comment back on those with actual numbers showing that, you know, for example, plumber, a software developer, median income in the US, a software developer gets paid over double what a plumber does.
• 40:09
  SPEAKER_01: There are about eight times more open jobs per year for software developers versus plumbers.
• 40:15
  SPEAKER_01: And the plumber field is growing by about 4% versus 16% for software developers.
• 40:21
  SPEAKER_01: So when you look at it with actual numbers, you go, well, that's a no-brainer.
• 40:26
  SPEAKER_01: But when you look at it with fear in your heart and just look at it with no actual information, just gut feeling, then it makes sense, right?
• 40:35
  SPEAKER_01: It makes sense to be a plumber because they're not being replaced by AI.
• 40:39
  SPEAKER_01: Neither are we.
• 40:40
  SPEAKER_01: So be careful you don't follow fear.
• 40:43
  SPEAKER_01: So let's look at actual information.
• 40:46
  SPEAKER_01: This first one is probably gonna be depressing, but we're gonna get to this.
• 40:50
  SPEAKER_01: Um right now, this is about US, right?
• 40:53
  SPEAKER_01: So this is the job postings on Indeed.
• 40:56
  SPEAKER_01: Just one service indeed, um, not every service.
• 41:00
  SPEAKER_01: So it's not how you represent them, it's not how you statistically exactly the same across every one of them, but job postings for the past year on Indeed overall.
• 41:10
  SPEAKER_01: This trend line points down into the right.
• 41:13
  SPEAKER_01: That's not a good trend line.
• 41:16
  SPEAKER_01: What this is showing us is in general, again, in the US, there's less jobs today than there were a year ago, or at least less job openings.
• 41:28
  SPEAKER_01: So when you look at software developers, you say, wow, there's less jobs than there were a year ago.
• 41:36
  SPEAKER_01: Guess what?
• 41:37
  SPEAKER_01: We're not alone.
• 41:38
  SPEAKER_01: We're not unique, we're not special in that way.
• 41:41
  SPEAKER_01: Because there are less jobs everywhere in the US, job openings now versus a year ago.
• 41:50
  SPEAKER_01: But I want to show you that yes, we kind of are special.
• 41:54
  SPEAKER_01: Because let's look at just software developer jobs, not all jobs in the US.
• 42:00
  SPEAKER_01: It's a different graph.
• 42:02
  SPEAKER_01: And in fact, it's up and to the right.
• 42:05
  SPEAKER_01: It's growing.
• 42:07
  SPEAKER_01: The number of jobs that are open right now is higher than it has been in a year.
• 42:15
  SPEAKER_01: We have more job openings for software developers today than we did a year ago.
• 42:22
  SPEAKER_01: And that is a continuing trend line we're seeing.
• 42:25
  SPEAKER_01: Now, yes, there's ups and downs with every trend line.
• 42:28
  SPEAKER_01: You can see that it's going up and down a little bit, but it's it's consistently going up and to the right.
• 42:34
  SPEAKER_01: This is a good thing.
• 42:36
  SPEAKER_01: I think this is what I've been trying to say all along is that when you look at it with a state of fear, with this idea that I saw this and that makes me scared.
• 42:48
  SPEAKER_01: Well, you're gonna miss what's actually happening.
• 42:52
  SPEAKER_01: When you hear a CEO say, we're laying off 4,000 people because of AI, there's another one that just happened recently where they laid off uh, at the end, I think 4,000 people because you know what, we don't need them.
• 43:05
  SPEAKER_01: We have redundancies because of AI.
• 43:07
  SPEAKER_01: And you find out, well, that's not actually the real story.
• 43:10
  SPEAKER_01: But when you hear those things, it makes you scared.
• 43:14
  SPEAKER_01: And when you see people using AI everywhere, and you think, oh, they're doing things that I used to do, but now the AI is doing it in minutes where I would take days to do it, that makes you scared.
• 43:29
  SPEAKER_01: But the reality, the actual data, is showing us that AI does not do a good job as a software developer, that software developers are desperately needed, and the number of jobs for software developers is growing even as the rest of the jobs in the economy seem to be softening or going away.
• 43:53
  SPEAKER_01: So while our economy in the US is on a downturn seemingly, especially In terms of jobs, software developers are still going up.
• 44:05
  SPEAKER_01: We need more of them.
• 44:08
  SPEAKER_01: So this is showing us that we are in a great field.
• 44:14
  SPEAKER_01: There's lots of reason for hope.
• 44:16
  SPEAKER_01: And that all this hype and all this fear that's been generated around AI isn't actually based in fact.
• 44:24
  SPEAKER_01: It's based in feeling.
• 44:26
  SPEAKER_01: And feeling isn't what actually long-term impacts anything.
• 44:33
  SPEAKER_01: Yes, feelings have impacted the software development job market for a long time now because people feel like AI is going to impact it.
• 44:42
  SPEAKER_01: And even CEOs are saying, hey, I could save some money because I feel like AI can do the job of software developers or customer service or sales or whoever else.
• 44:53
  SPEAKER_01: But when they actually try and do it, they find out they're wrong.
• 44:56
  SPEAKER_01: And that's their feelings.
• 44:58
  SPEAKER_01: Well, that's how they felt, it's not what reality is.
• 45:02
  SPEAKER_01: And because of this, we're seeing a direct impact on the software development field.
• 45:10
  SPEAKER_01: We're seeing growth in the field.
• 45:12
  SPEAKER_01: We're seeing more people that need software developers to do the actual work, to utilize the AIs and corral them so they don't have as many bugs going into production.
• 45:24
  SPEAKER_01: So they can keep that down so I can still create software because software is desperately needed.
• 45:30
  SPEAKER_01: And yes, they're hiring us to fix the problems that they created.
• 45:35
  SPEAKER_01: So, with that being said, what's the conclusion of all this?
• 45:40
  SPEAKER_01: So the conclusion is I think you should be very hopeful for software development because the future is bright, but it isn't going to be bright for everyone.
• 45:50
  SPEAKER_01: And this is one of those things that I have learned as I talk to people is that they think that everyone, if they say they're a software developer, should either have a job if it's a good economy, or they shouldn't if it's a bad.
• 46:06
  SPEAKER_01: And so if they don't get a job, it must mean that it's a bad economy because they aren't getting a job.
• 46:10
  SPEAKER_01: And they've applied for a month through even a year.
• 46:14
  SPEAKER_01: And as we saw from a job posting on Indeed, software developers, the need is going up, but the feeding frenzy from the COVID era is gone.
• 46:24
  SPEAKER_01: In fact, let's go back to that graph for a minute.
• 46:27
  SPEAKER_01: I'm gonna change this timescale from one year to 10 years, which actually this graph only goes back to like 2020.
• 46:33
  SPEAKER_01: So like six years.
• 46:36
  SPEAKER_01: But if we look at this graph now, we can see the lower right-hand corner, there's our little green diet line, and it's going up.
• 46:44
  SPEAKER_01: But if you look at what came before, you're like, well, but that massive peak in 2022 is so much higher than where we are today that it feels like you know, it's not even close.
• 46:57
  SPEAKER_01: And in fact, we're just kind of skimming along the bottom here.
• 47:00
  SPEAKER_01: Why is that?
• 47:01
  SPEAKER_01: Well, the reason for that is because during COVID, when the interest rates were practically zero and there was a whole bunch of free money floating around there, and companies were also thinking that everything's gonna go digital, and we had to do more uh app delivery things and you know like remote work and all sorts of stuff, it required a whole different set of software.
• 47:25
  SPEAKER_01: And so big companies, small companies, everyone was gobbling up anybody that said they were a software developer.
• 47:33
  SPEAKER_01: And that created this huge boom of people being hired.
• 47:39
  SPEAKER_01: And if you had just gone through a few tutorials and had a little bit to show, you probably got a job as a software developer.
• 47:47
  SPEAKER_01: But that era from COVID is gone.
• 47:50
  SPEAKER_01: And in fact, most those people got laid back off because they weren't needed after all.
• 47:57
  SPEAKER_01: And those companies got through COVID and now are back to you know, quote unquote normal.
• 48:02
  SPEAKER_01: And all of a sudden, we now have this era where we're back to where we used to be.
• 48:09
  SPEAKER_01: So the COVID era is gone.
• 48:12
  SPEAKER_01: Gone are the days you just go through a few tutorials and get that job.
• 48:16
  SPEAKER_01: So, what is it gonna take in order to get a job?
• 48:20
  SPEAKER_01: Because there's a lot of people that are out there saying they're software developers who aren't getting work.
• 48:26
  SPEAKER_01: What it takes is work.
• 48:29
  SPEAKER_01: And let's talk about those steps to take full advantage of this Russia software developer because it's not gonna be just any software developer that gets a job.
• 48:39
  SPEAKER_01: It's not gonna be just, well, I've got 20 years experience.
• 48:43
  SPEAKER_01: It doesn't matter because that's not what's important to companies.
• 48:48
  SPEAKER_01: Yes, they say we want X number of years of soft of experience, but what they're really saying is we want you to have a depth of skill.
• 48:57
  SPEAKER_01: And if you don't have a depth of skill in current technologies, you're gonna struggle.
• 49:04
  SPEAKER_01: So, number one, I want you to know more than just how to use an AI.
• 49:09
  SPEAKER_01: Again, AIs create errors at about 1.7 times that of software developers.
• 49:16
  SPEAKER_01: So if you are utilizing an AI and not knowing how to actually do the work and check the work and know if it's right or wrong, then you're useless.
• 49:27
  SPEAKER_01: And I'm sorry that sounds harsh, but you're gonna why why not just let you know OpenClaw or a Ralph loop or something else do the work of that AI and cut you out of the loop.
• 49:40
  SPEAKER_01: And that's what's gonna happen.
• 49:42
  SPEAKER_01: You will be laid off because you don't provide value.
• 49:45
  SPEAKER_01: If you can't control the AI, if you can't tell the AI you're wrong, then you don't provide enough value to be a software developer.
• 49:56
  SPEAKER_01: So you need to be able to do more than the AI does.
• 50:01
  SPEAKER_01: You need to know more than just how to use an AI.
• 50:06
  SPEAKER_01: That's not the marketable skill that you think it might be.
• 50:10
  SPEAKER_01: Instead, the marketable skill is being able to say, yes, that code runs, but it's not secure.
• 50:17
  SPEAKER_01: Yes, that code runs, but it's not performant.
• 50:20
  SPEAKER_01: Yes, that code runs, but it's not the way we should be doing it.
• 50:24
  SPEAKER_01: If you can't do that, then you aren't providing value to an employer.
• 50:30
  SPEAKER_01: Number two, you need to have a depth of knowledge, not just a basic understanding.
• 50:36
  SPEAKER_01: Because you know what?
• 50:37
  SPEAKER_01: The AI has a basic understanding.
• 50:39
  SPEAKER_01: The AI has a very good understanding of every language.
• 50:44
  SPEAKER_01: So if you say, well, I built a few things in this or that, it's not good enough.
• 50:50
  SPEAKER_01: You need to have a depth of knowledge.
• 50:52
  SPEAKER_01: You need to be able to go beyond just, I built this little thing, because that's what the AI does.
• 50:57
  SPEAKER_01: What you need to provide is, again, that depth of skill to say, I know how this should be done.
• 51:04
  SPEAKER_01: And I know there's five different choices here, and I know that's the wrong choice.
• 51:09
  SPEAKER_01: So you need to know about those other ones, which means it's not just about knowing syntax.
• 51:15
  SPEAKER_01: The AI knows all the syntax.
• 51:17
  SPEAKER_01: What you need to know is how to apply that.
• 51:21
  SPEAKER_01: Have the wisdom of having built applications so you know this works here and this works there, and they shouldn't be mixed.
• 51:30
  SPEAKER_01: Number three, you should be well-rounded.
• 51:33
  SPEAKER_01: You shouldn't focus on, well, I'm just a desktop C sharp developer, and that's it.
• 51:39
  SPEAKER_01: That's going to leave you with less options.
• 51:43
  SPEAKER_01: You need to be well-rounded.
• 51:45
  SPEAKER_01: You need to be able to understand about DevOps and about the cloud, about security, about efficiency, about working with a database, and so on.
• 51:53
  SPEAKER_01: Why?
• 51:54
  SPEAKER_01: Because AI is opening up new opportunities for developers.
• 51:59
  SPEAKER_01: But a lot of those opportunities are in small to mid-sized companies where they wouldn't have or couldn't have afforded a larger software development team, which means you'll have to wear more hats.
• 52:12
  SPEAKER_01: But you'll be working in a company that didn't have a job opening at all a year ago or two years ago because of AI.
• 52:20
  SPEAKER_01: But you need to be well-rounded so you can make wise decisions all the way up and down the supply chain for software.
• 52:28
  SPEAKER_01: Number four, you need to have real experience.
• 52:32
  SPEAKER_01: Too often I see people that have gone through a hand-holding exercise, I'd call it, where I do this with my tutorials too, where I will show you how to do something step by step.
• 52:45
  SPEAKER_01: That is valuable.
• 52:46
  SPEAKER_01: I am not discounting that.
• 52:48
  SPEAKER_01: But do not think that you now have the experience to do that exact same thing because you watched me do it and followed along.
• 52:58
  SPEAKER_01: That's a hand-holding exercise.
• 53:01
  SPEAKER_01: And those are valuable.
• 53:02
  SPEAKER_01: But just like with a toddler who's growing up, yes, they hold on to their parents' hand, they can more easily walk.
• 53:09
  SPEAKER_01: But at some point, you have to let go of the hand and walk on your own and then learn to fall down and pick yourself back up.
• 53:16
  SPEAKER_01: That's what this is when you're learning software development.
• 53:20
  SPEAKER_01: Yes, hold the hand at first, but you have to go off and start building things on your own, real things.
• 53:27
  SPEAKER_01: Number five, show off what you can do and why it's valuable.
• 53:32
  SPEAKER_01: If you're not showing off what you can do, how do people know that you can do it?
• 53:37
  SPEAKER_01: You can't just say it.
• 53:39
  SPEAKER_01: There's too many people out there that are saying what they can do.
• 53:43
  SPEAKER_01: You need to show it.
• 53:45
  SPEAKER_01: Otherwise, you're just another voice in the crowd.
• 53:49
  SPEAKER_01: Okay, so those are five things, five ways to stand out and take advantage of this new wave of incoming software development jobs.
• 54:00
  SPEAKER_01: You need to become more than just an AI jockey.
• 54:02
  SPEAKER_01: Be the adult in the room that can make informed decisions that will impact the software and the organization for years to come.
• 54:10
  SPEAKER_01: That's how you become valuable in today's marketplace.
• 54:13
  SPEAKER_01: Most jobs are shrinking right now.
• 54:16
  SPEAKER_01: And while software development is growing, you still need to be prepared to make an impact.
• 54:21
  SPEAKER_01: And you need to be able to do that by showing off those skills first so you can show off I can do this for you.
• 54:28
  SPEAKER_01: All right?
• 54:28
  SPEAKER_01: So there is hope in software development.
• 54:31
  SPEAKER_01: This industry is growing.
• 54:33
  SPEAKER_01: We are gonna have more of an impact.
• 54:36
  SPEAKER_01: But we have to make sure we focus on facts, not just fear.
• 54:40
  SPEAKER_01: All right.
• 54:40
  SPEAKER_01: Thanks for watching.
• 54:41
  SPEAKER_01: Thanks for listening.
• 54:42
  SPEAKER_01: And as always, I am Tim Corey.
• 54:45
  SPEAKER_00: Thank you for joining us for this episode of Dev Questions.
• 54:48
  SPEAKER_00: When you're ready to learn to think and code like a professional developer, head over to IamtimCorey.com and enroll in a course.