Develop Yourself
To change careers and land your first job as a Software Engineer, you need more than just great software development skills - you need to develop yourself.
Welcome to the podcast that helps you develop your skills, your habits, your network and more, all in hopes of becoming a thriving Software Engineer.
Develop Yourself
#263 - They're Lying to You About AI Productivity: Hype vs Reality for Coders
Maybe I'm just coping.
Maybe.
I’ve been pretty vocal about the over-hype of AI coding tools and I always get the same responses from AI bros:
“YoU gOtta PrOmpt bEtteR!”
“This iz a skillz issue dawg”“I’m using [x] and it works perfectly”
“I’m a super duper senior architect and I’ve replaced my entire team with AI agents. You’re coping.”
I’d be lying if I said I didn’t doubt myself.
Maybe they’re right, maybe I’m just not using these tools correctly.
Let’s take a look at some big fax and small fax to understand where the truth ends and the hype begins.
Sources:
TrueUp Software Engineering Job Trends (June 2025)
Business Insider: Google engineers just 10% more productive with AI (June 2025)
Computerworld: Minimal productivity gains from AI chatbots
TechRepublic: IBM Study on AI ROI (2025)
https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/
Shameless Plugs
🧑💻 Join Parsity - For career changers who want to pivot into software.
✉️ Got a question you want answered on the pod? Drop it here
Zubin's LinkedIn (ex-lawyer, former Googler, Brian-look-a-like)
Welcome to the Develop Yourself podcast, where we teach you everything you need to land your first job as a software developer by learning to develop yourself, your skills, your network and more. I'm Brian, your host. I was recently talking about my experience using AI tools for coding and then somebody told me I was just coping and I was like, am I? So? I've been pretty vocal about the overhype of AI coding tools and I always get the same responses from these AI bros. They go something like you just need to prompt better. It's a skills issue. I'm using Claude or Gemini or Grok or whatever, and it works perfectly. Or my favorite one, I'm a super duper senior architect and I've replaced my entire team with AI agents and you're just coping.
Speaker 1:I'd be lying if I said I didn't doubt my own experience using these AI tools and sometimes I even wonder maybe they're right, maybe I'm not just using them correctly. But now I'm convinced that these people are either liars, they are naive, or truly top 0.01% coders with whom mortals will never be able to compete. If you were already a 10x engineer like seriously in the top 0.01% or 0.5% of engineers out there and you're using AI tools, I have no doubt that you're probably incredibly more efficient than you ever were or more than most people could ever be, but it's getting very annoying because every other day I'm reading about some developer on LinkedIn or X or even YouTube who's 10Xing their output with AI tools. And then you have companies like Salesforce or Microsoft that are claiming that AI is making them so much more productive that they just have to fire developers. I want to explore some facts, not feelings, to understand where the truth ends and the hype begins when it comes to AI tools, and to understand where we are now.
Speaker 1:It's important to look back around six months ago. Why six months ago? Does anybody remember about six months ago when our lizard king, mark Zuckerberg, told Joe Rogan that AI would be replacing mid-level engineers, or writing code at the level of a mid-level engineer in about six months? That was six months ago. Do you remember when Devon AI, a startup that was supposed to replace all software engineers, came out with that bold statement almost a year ago? And yet here we are.
Speaker 1:Despite all these catastrophic events, there are somehow more jobs right now for software developers in July of 2025 than in July of 2024, and more than in July of 2023. And I have the sources for all these stats in the show notes. These aren't just things I'm pulling out of my hat. I'm going off facts and I'm looking at a site called TrueUp, which I often cite on this show because it shows you up-to-date, real-time information about the number of tech jobs that are open. It's almost as if these CEOs are pulling predictions straight out of their hats. Yeah, that's what I wanted to say.
Speaker 1:Salesforce also, around six months ago, bragged that they were cutting a thousand jobs this year because of AI, but then they started hiring for more AI aligned roles like sales engineers. It seemed like they were basically shifting their head count, which leads me to one of the worst and most pervasive lies about AI is that it somehow made one senior developer equal 10 junior developers, or it's completely wiped out the need for us to hire a junior developer. I call BS on this. Ask any developer who's working on something moderately complex that includes multiple services, a back-end system that is more than three months old, how much using Cursor or Grok or Claude has helped them with their day-to-day workflow. It feels like we're all a little bit afraid to admit the truth, because the truth is it depends Depends what you're working on, how old it is the level of complexity, the level of code quality and your own skill. It's just too hard to say how much it really improves our productivity. Meanwhile, you have these geniuses on LinkedIn that tell you they've effectively just replaced an entire team, or at least that they could if they wanted to. That's how effective these tools are from their point of view. They said they have an MCP server, they have these beautiful prompts and rules files that go through every conceivable scenario that a coding agent may encounter, and they write prompts like a cold dead robotic Shakespeare. And if you're not seeing those productivity gains, you're just using it wrong, bro.
Speaker 1:Well, google did a little study on AI and its impact on productivity. What did they find? 10% more effective is how much they measured their engineers were with AI assistance. Again, that is coming from Google, who has a lot of incentive to inflate these numbers, and they also are well known for meticulously documenting these kind of statistics and metrics. I don't know why they would make up a lie that would make the AI tools look worse. 10%, by the way, is incredible. I mean that's an incredible amount more. You think about a billion dollar company outputting 10 more percent. If that can translate to their overall profit. That's amazing If they can reduce their headcount by 10%. Still amazing Is this 10X. It's not even anywhere close.
Speaker 1:Now, outside of tech, these gains are more modest. They're barely there. One study found that office workers unlocked about 3% more productivity. Now I get it. We're in the beginning stages of using these tools, but I think that the hype is really outpacing the reality by a long, long shot. And if you're a person using these tools, I would ask you how do you actually feel using them? Are they actually truly making your job basically on autopilot? Are you using agents to decide what to do on a daily basis? Have you had an agent go on a website, buy you a ticket to Florida or California or Texas or Disneyland or whatever, negotiate your ticket prices, send off emails and basically do everything you wanted to do?
Speaker 1:I noticed that nobody's made that AI agent yet. Why are all the AI agents so focused on writing code? Could it be that code is repeatable? There's tons of data on the web about how to write code and it follows really clear patterns for the most part, and even then we're making 10% productivity gains. When you let an AI agent run wild, they don't seem to work. There's another study that I didn't cite here, but I read this morning that says about 70% of the time, agents are essentially like an autonomous AI tool that just goes out kind of give it a generic, nebulous, open-ended task and it just does it.
Speaker 1:I actually did a show earlier on this about Anthropic and this model called Claude and they made this little Claudius bot that could run a small business and it failed miserably. It had hallucinations and it even thought it was an actual human being wearing a nice little red tie and said it was gonna show up at 742 Evergreen Terrace, which is the Simpsons' house. So yeah, I don't think we're quite there yet, but why ruin the fun with facts and numbers from independent researchers and massive tech companies who meticulously track these kinds of stats? That wouldn't make for very good YouTube videos, would it? Hey, I hope you're enjoying this episode Now you know that I own an anti boot camp with my buddy Zubin, an ex Google software engineer.
Speaker 1:If you're interested in not just learning how to code and you know it's going to take more than three months and you're serious about making a transition into a career in software and you want to work with people that have done it before and are currently working in senior plus levels. Join me and Zubin at parsityio slash inner dash circle. You can learn all about our philosophy, how we approach learning how to code and switching careers in a much different way, and how we have so much gosh dang success. If you're interested in being one of the few people that works with us this year, go and apply at parsityio slash inner dash circle. And now back to the episode.
Speaker 1:And finally, the big elephant in the room is that the ROI is just not there quite yet on these tools. Roi means the return on investment. Really, what it feels like it means is really over. Indexed Companies weren't just happy enough with like the magical powers we've gained from having AI tools like Cursor, copilot, chatgpt. These tools are amazing. I'm not going to lie and say they're not. I use them all day long, all the time, and I pay for them. I don't hate them at all, but these companies wanted earth-shattering, society-crippling change. Autocomplete on steroids just doesn't have the same ring as AGI or LAOS.
Speaker 1:So these people wrote slide decks promising the world to investors who have already been burned by poor planning, high interest rates, bets on Web3, blockchain and crypto. How are those NFT, web3, and crypto investments doing, by the way, probably not the best right. Ibm did a survey on thousands of executives and found that only a quarter of AI initiatives have achieved the respective return on investment so far and only 16% have successfully scaled AI across the enterprise. To many people, this would be considered a failure, but when it comes to AI, the idea is you just got to wait. Just wait another six months and we'll see those bets paying off for real. This time and I get it it's better to invest in a new technology and be burned than not invest and be late to the party or miss out on an explosion of growth and opportunity aka making a bunch of money. But I do think it's interesting how optimistic all these executives are, despite not seeing any return on investment yet. If this was any other technology, I don't believe we'd be seeing the same optimism Now for every study I've referenced here.
Speaker 1:There's probably some other study that's going to show the opposite, and those studies, from what I see, are almost exclusively written by people who have a lot to gain from AI hype. In all honesty, I have a lot to gain as well. I'm going to be completely transparent here. I'm a software engineer and I also own a business called Parsity that mentors early career developers. How much can you really trust anything? I say so. I have a really novel idea for you. Use the tools yourself. Use them on a non-trivial project with a team of more than three people in a code base that isn't exclusively front end. Draw your own conclusions. Read studies, look at reality.
Speaker 1:Do you personally know a single person who has been replaced by AI? I used to ask a similar question to people who are really into crypto. I'd say do you know anybody who's bought a slice of pizza with crypto? At a party a few years ago, somebody was telling me how we won't be needing money in a year or two, and I thought I don't think so. Common sense shows me. All our collective experience shows us that no one's really using crypto right. So the idea that overnight it would just be widely adopted by the whole world and we go into this new form of currency made no sense. And you could do the pizza test have you ever bought a slice of pizza with crypto? Do you know anybody who's bought anything with crypto? Most people are going to say no, and the same thing happens if you ask do you know a person who's been replaced by AI, like, literally, there's an agent doing their job and there's a lot of arguments to say, well, I wasn't quite replaced, but my job was made redundant because of AI. And I would argue back that a lot of these layoffs are not the result of AI, but the result of overhiring during the pandemic and companies feeling the burn and the pressure from investors to have more money and do more with less, because interest rates have sustained a high that we haven't seen in our lifetimes.
Speaker 1:I don't want to make it seem all doom and gloom, though right, because everything is not bad. Personally, this is the most fun I've had coding so far in my career, and the reality is that I am more productive, I don't have to write nearly as much code and I can build a web app in minutes instead of days a really small, simple one, or a prototype of one. I love these tools. I just hate the hype and I hate what they're doing to people that are thinking of breaking into the industry or trying to get their first job and thinking that the world is conspiring against them. I don't believe that's the case. I do believe there's actually a really strong future in software development in general, and I think that these tools are scaring away a generation of people that should be in this field. That may not because they're scared. These tools may wipe away all the jobs, and I just don't see it.
Speaker 1:I'm sure there's people out there that are smarter than me that are going to claim that we're just a few more months away from these things, but right now we've been waiting for months and months and months, and the technology will not get exponentially better. That's the last thing I wanna end off with. There's this weird pervasive claim that, oh yeah, it's just gonna get better next year and then exponentially get better. But looking at any technology, we know that's not the case. Space travel hasn't exponentially improved since the 1960s. Neither have cars, neither have homes or ovens or other technology we use every day. In fact, even phones have basically plateaued. There's a lot of money and a lot of geniuses working on these problems, so to think that AI would somehow just buck this trend and instead get exponentially better every single year until forever doesn't make any sense. It's not based in any reality.
Speaker 1:So again, take what I say with a grain of salt. Use these tools for yourself. Maybe don't get caught up in all the AI hype online and then make your own conclusions based on your experience and what you read in SPAC's research reports, rather than just really loud voices on YouTube or LinkedIn. Hope that's helpful and I'll see you around. That'll do it for today's episode of the Develop Yourself podcast. If you're serious about switching careers and becoming a software developer and building complex software and want to work directly with me and my team, go to parsityio, and if you want more information, feel free to schedule a chat by just clicking the link in the show notes. See you next week.
