When a Technology Removes Scarcity, Some People Defend the Scarcity as Though it Were Quality.

The lack of evidence Does not equal quality, When a technology removes scarcity, some people defend the scarcity as though it were quality.

There are a lot of people who make their identity their job. Their job made them feel secure. The security gave them a sense of self-worth, and a lot of that now is starting to disappear. You spend years honing your skill, mastering it, and one day it gets up-ended and you have to defend it. Defending it is defending your very identity itself. Your skill being obsolete is now your self being obsolete. It’s a hard thing to cope with. You were one of the few people that were able to do this skill. You earned good money, and now it’s just gone, almost vanished overnight.

The failure to adapt to new situations that arise in the world, it can be very difficult. A lot of people don’t even see it as a failure to adapt. They see it as “this is still relevant, and I’m going to defend it with my life.” They have yet to see the writing on the wall. Maybe because it’s in its early stages, but it’s happening, whether they like it or not. The best comparison I can think of when we talk about AI and software development is the horse and buggy versus the car. The whole industry made around maintenance of the buggy, taking care of horses, places to put your horses on the side of the road, and the cleaning up poop off the road. That was shattered almost overnight because the car was made. It was more convenient. It was more efficient. It’s cheaper. You don’t have to feed it, you don’t have to give it water, you dont have to train it, clean it, pick up after it. It just does the job of getting you from point A to point B. Yes, there are some nuances. Things break on cars, there’s maintenance you have to do on them. We all understand this, but overall, it’s faster. It works better, and it’s less maintenance, and cheaper. People who love horses still own horses. They ride them, they race them, they take care of them. It’s more of a sport at this point, a hobby per se, an animal lovers crux. In other words, they do it for the love of horses because they like horses. They can still interact with horses. Nobody has a problem with that.

I see the same thing happening with software developers who refuse to use AI for any type of development. Not to say these guys don’t have skills. We still need software developers that understand and can manually hack it. We need those guys to build the AI to code, the smartest of the smart people you would say. Let’s be real: not everybody can do that. There’s probably, at this point, 0.001% of the population maybe even more that can actually operate at that level. They’re already working for OpenAI, Claude, Perplexity, Google, Microsoft, Chinese companies, Huawei, and others. Right now, what’s going on is the horse-and-buggy people who have jobs, a skill that pays their bills, aka the software engineers, who refuse to switch over from a horse and buggy to, say, a mechanic maintaining engines and learning about Cars. They are arguing the horse and buggy is superior and better in all ways, and that the car is inefficient. It breaks down too often, and it doesn’t get the job done as well as the horse and buggy Because of those things. AI is sort of the same. It’s in its early stages. It hallucinates. It’s not perfect. AI was built and trained off of humans. Humans aren’t perfect. Take a look at the junior to mid-level engineers. They’re not perfect. They could write slop code. They can write security issues Into their code as well and do. To expect this brand new technology to come out working perfectly with no hallucinations, don’t make any mistakes, be perfect, no security vulnerabilities, design perfectly, it’s just not realistic. It will get better over time, just as the car got better over time, to where nobody even questions whether the car is better than the horse or the horse is better than the car. It’s obvious. In the future, in my opinion, the software developer that hand jams everything will be looked at as maybe an artist, or doing it for the love of just coding and solving problems. They might discover great new things, but it doesn’t take away from the fact that AI can do it ten times faster. At this point, Fable 5, Sol, Kimmy K3, and Quinn are pretty accurate, and they’re getting better and better every day.

Ability to change your mind or adapt is a form of competence. It’s recognizing that things are changing, nothing ever stays the same for long, and that you’ll have to do something different. If you’re a technology professional, you like technology. This is the new technology. It’s AI. I don’t know what else to tell you, so if you don’t like it, you’re going to get left behind in technology. Back in 2021 or something around that time, when ChatGPT first came out, I thought it was great. I was able to use it to look things up faster than I would using Google. I kind of used it as a Google replacement back then.

I had my own coding projects, so it was able to help me with it. I would copy and paste it into the chat. The chat would analyze it and then give hints or better ways to do things, and then start giving me code that I can implement or make better. I copied that from the chat, put it into my IDE, tested it, made sure it worked, clicked through it, and vice versa. It was a slower process. While going through that process, I thought, “Wow, wouldn’t it be great if I could get this onto my local computer and just have it do it, so I don’t have to copy and paste, and then I could just review what it proposed and implemented?” So, in other words, when it started out, it was helpful for me to do quicker research to synthesize articles that I was reading and maybe want to do research on and get the bulk of the information out to write papers myself. I was going through college at this time. It wasn’t changing my entire workflow. I still have my research process. I still have my writing process, I still review and read everything, code or otherwise, Still programming by hand but not great at it lol. It just sped it up at a quicker rate and gave me things I would not be able to do because I was not great at codeing but empowered me to do so anyway. Instead of googling, I just “ChatGPT’d” it, and yeah, sometimes it was wrong. You had to correct it. So I tried to experiment with an open-source model on my computer. It basically pegged everything to the top (ram, cpu, ect.). It was slow. It was able to read some files, but the inference wasn’t right. It had a small context window, and it wasn’t very good, this was back in 2022 I think. I kind of just completely ditched it and stuck to mainstream models. I stuck with it using AI over time. It got better and better and better and better. Was it November? About November 2025, Claude Opus 4.8 came out with agents, and it was able to manipulate the things on your computer and spin up swarms and perform adversarial reviews, just like I wanted it to Since the beginning of 2021. At this point, I was already keen to understand the downfalls of AI:

• Where it hallucinates

• How it could get things wrong

• Context windows

• When to restart your context window

• When it compresses conversations a couple times, you need to start a brand new chat because it will lose important details when compressing

• The rabbit holes that it would fall down and to pay attention to those so it doesn’t waste your token usage

• Make sure you have a clear direction on what you’re trying to accomplish

• Research and understand the infrastructure you’re putting it on.

• Program with a security mindset first.

• Just things that a general programmer, cybersecurity professional, or just IT technology professional would think of when they’re implementing a system.

Just things to make AI work better on my side

So, in turn, I was taking all my stored-up knowledge that I learned manually. I researched, I hand jammed, I reviewed, and I spent countless nights over it, and it started to speed my process up ten times ×30 ×40 ×50 even. The way my brain works and my arch-type, this is great! I love it, but for people who get the satisfaction from arranging, say, a logical construct manually, they may have a difficult time with this transition. It’s taking that away from them. It might not be exactly how they wanted to do it, even though it works. The mechanical work in other words is being automated. Someone who is primarily motivated by making products or loves the satisfaction of seeing their idea in that they imagined in their head come to life and work all of a sudden has the ability to do more than they ever could before. So I guess a person who thinks “I’m valuable” because I know how to manually perform this difficult process or task and solve this logical problem will experience, ultimately, an erasure by automation. A person who thinks I’m valuable because I understand the problem can orchestrate how to solve the problem with AI and not get slop out of the outcome because I know how things connect. I know how things work. I just don’t know the tiny little nuances like where “semicolon placement is supposed to land or the space that was missed”, but I understand the problem as a whole and now have empowerment, People who think like this, can now leverage this.

Okay, so I’m going to compare this to a security analyst who sees their identity as the equivalent to manually enriching an alert. They’re going to feel threatened By automation they are going to make up every excuse to why it’s bad. But the security analyst who sees their identity in maybe understanding adversaries, making containment decisions based on the incident that happened, or protecting operations will see automation as a way where they can just stop wasting their time on little things that, in the grand scheme of things, don’t matter. Now, developers, grapphic designers, and lawyer examples:

• A developer who identifies with typing code will feel replaced.

• A developer who identifies with designing systems will gain a larger design surface.

• A graphic designer who identifies with controlling Photoshop may resent image generation.

• A designer who identifies with communicating ideas visually may add another tool.

• A lawyer who identifies with producing standard language may feel threatened by document augmentation.

• A lawyer who identifies with judgment, negotiation, interpretation, and accountability may use automation to remove lower-value work.

I think the overall battle that’s happening can kind of be summed up like this right now. When the method is your identity, automation feels like you are being destroyed. When the outcome is your identity, automation feels like you have the power to do anything. This is what AI allows and is causing. I think a lot of these super advanced and experienced software engineers are really missing out on the way that AI can empower them. They have the expertise; they know almost all of it. They can basically make anything. They made the internet at this point; they made AI, but they’re getting too caught up on “oh, they’re taking my ability away to type on a keyboard.” It honestly sounds ridiculous. These guys could be so great if they embraced it they could even made it better. I know a lot of them are starting to come around, but there’s a big pushback on the internet that disagrees with this, and they hate AI, and there’s some weird cult-like behavior around not using AI and hating it and doing everything you possibly can to discredit it. Same on the other side there are also people who exhibit cult like behaviour about AI. Now, don’t get me wrong: AI has the ability to do horrible things, not just for society but for people personally. There’s things called AI psychosis, Anthropic is now doing watermarks in the text, dont be surprised if other companies like open AI, Google and Microsoft start doing the same thing. In my opinion, that’s more dangerous than the AI itself. Its kind of like putting a stamp of ownership on every single piece of intellectual property that has run through “The AI” Basically claiming the “AI company” is the one who came up with it and now owns it by putting their watermark on it. It says that it went through their system, so now they own it and it was AI generated. Think about this: AI is starting to be restricted by the government or regulated by the government. Now you have watermarks. You’ve run some of your stuff through AI just to check it. It got watermarked. You didn’t know it got watermarked. You can’t see the watermark, but now you’re being taxed because there is a new law on the books that taxes AI generated things. There’s a whole bunch of crazy stuff that could happen, right? I’m not saying it’s all good.

What I’m saying is expertise is becoming more valuable at the level of judgment and less valuable at the level of repetitive execution. I’m predicting a lot more smaller organizations will pop up with less bureaucracy to bottleneck implementation and how quickly things can move towards production. For example, the big older established companies have their processes set based on older systems. There are internal layers they have to be worked around. It’s optimized for that type of world. Here’s an example based on maybe a cybersecurity thing / issue holding up production and why these bigger organizations probably ultimately won’t be able to keep up. Although cybersecurity is very important, don’t get me wrong: I’m not discrediting that you dont need it, but security can become one or more approval layers that exist primarily to preserve its own authority.

With nuance, a proposal enters security review. Security asks for another document. The document creates another meeting. The meeting produces a committee. The committee refers the question to legal. Legal asks security to define the risk. Security says it is not comfortable. Six months later, nobody has designed an actual control. The company has used the existence of a process with the management risk, and nothing happens. Nothing gets done. Security people are just “no people”.

AIn the earlier AI error there was some type of delusion that people had. I could admit that there’s an enormous amount of different claims saying, “I make $10,000 a month doing this with Claude or ChatGPT, this, that, or the other,” and lots of scams going on. Not an enormous amount of convincing work, the difference now is that working systems are actually being shipped. I myself have a couple of working systems:

• One is about weeks away fro launch

• The other is in beta testing.

• One is an iOS app.

• The other is an open-source security operations center that I’m still working on, but it’s downloadable.

• I’ve built working phone systems with Twilio For super cheap To where I have it connected to an IP phone and I can call out from multiple different numbers that I own from my cell phone

• I built real business automation that would otherwise cost me an enormous amount of money and time. I’m doing it for not even a fraction of what it would cost me.

Another thing that’s damaging to the AI community is that there’s a lot of overexaggeration: calling everything agentic, generating piles of low-quality output, and pretending the existence of code is the same as a working product. It’s not true. There’s a lot of nuance in there, and it’s harder than people make it out to be. You can’t just vibe code a SaaS application. There’s a lot more that goes into it. Those people that say those things and make it seem like everything is way easier than it is, it’s just fueling the fire and giving the critics of AI easy targets to point at and poke at justfuly so. When you create something, you still have to follow testing rules.

• Test the current tools

• Give them real work

• Measure results

• Inspect the failures

• Add controls

• Ship something useful

Okay, we’re going to keep harping on the security, right? It’s the clearest test of whether someone’s a builder or a blocker. I guess you could say AI systems are becoming extremely capable at identifying combinations and vulnerabilities that would be difficult for an individual human to discover, or, if not, someone who doesn’t really have experience in that aspect could greatly benefit from them. I also recognize that critical safety systems do require more careful inspection than ordinary software itself. You can’t trust exactly what the AI said, you should trust but verify. So if you’re a security professional or any other type of professional and you’re using AI, there are some things you should probably ask. Ask thes questions instead of saying, “Don’t use it.”:

• What information will enter the system?

• Where will that information be processed?

• What systems can the agent access?- What actions can it perform autonomously?

• What requires human approval?

• How are actions logged?

• How are generated changes reviewed?

• What is the blast radius if it behaves incorrectly?

• Can the action be reversed?

• What happens when the model or provider changes?

• What is the incident response plan?

That is security engineering. And yeah, sometimes the final answer will still have to be no, the particular model integration, data flow, or autonomous action may create unacceptable risk But it should be produced from a Conclusion based on analysis Not just “don’t use AI, its not secure”

This long-winded article concludes: At the end of the day, I’m not saying AI is perfect. I’m not saying every company should hand over production access, customer data, or its critical infrastructure to a model and just hope for the best. I’m saying that fear is not a strategy, nostalgia is Not quality. Protecting an old workflow is not the same thing as protecting society. I get it. I understand why people are angry. To spend years of your life mastering a skill you could build a career around, make a solid life, and make it part of your identity, and then watch the skill basically more or less become useless overnight. I have a Network Security Bachelor’s Degree and a cybersecurity master’s degree. I have almost 11 years in the Navy as a submarine sonar tech, with work experience leading people, two years as a SOC analyst at VMware, I feel it too, and I’ve already felt it, I lost my job and am struggling to find a new one. You’re allowed to be uncomfortable with that. You’re allowed to grieve the old ways of doing things, but grief can’t become a permanent veto of progress. I love and hate AI, I think the people who are going to get through this and survive this transition are not necessarily the best or most experienced people.

They are not the people who can type the fastest or memorize the most syntax, manually perform the most repetitive tasks, or say no with the most authority. They’re gonna be the people who understand the problem as a whole, ask the right questions, recognize when output is wrong, put controls around it, and still find a way to move the mission forward. That’s where expertise is going. It’s moving from repetitive execution to judgment. Now that judgment, like I said earlier, needs a foundation. Keepers of access are going away. It’s going from being the person everyone has to go through to being the person who makes sure the final result actually works. Nothing wrong with loving the old craft of programming on the keyboard. Keep coding by hand if you want. Keep drawing on paper. Keep writing without AI. Great things, I think, will still come out of that. Keep riding horses! Do it because you love it. Do it because the process itself gives you satisfaction It will be uniquely human and satisfying. Just because you love it and you think it’s the end-all be-all doesn’t give you the right to force everyone else to stay there with you. The fact that something used to be difficult does not mean the difficulty itself was valuable. Scarcity does not automatically equal quality, and removing a bottleneck does not automatically destroy the profession. AI is not coming. It’s already here. The real question is not whether it will change your profession. The question is whether you’re willing to change with it. You can help steer the car, you can improve the engine, install the brakes, make it safer, and decide where it should go, or you can stand beside the road yelling that the horse was better. Either way, the car is already moving.