AI is the threat to every "Lập trình viên"

#1

Hi

I get the impression that GenZ is gradually becoming addicted to OpenAI and Anthropic. This is evident in how adept they are at using ChatGPT or Claude. This post Tạo landing page bằng AI Agent có thực sự hiệu quả không? confirms this and is just the tip of the iceberg.

This raises a provocative question: “Why should I spend at least four years at a university just to earn a degree in computer science when I can create ‘beautiful code’ with ChatGPT or Claude in just a few minutes?”

I myself find it hard to give you a satisfactory answer to that. Perhaps I’m someone who lives in the past, because I consider the idea of skipping a university education by learning ChatGPT or Claude to be absurd and harmful to a country.

The old saying “No pain, no gain” basically means that you won’t achieve anything if you don’t work hard. But this saying is now outdated and completely obsolete, because with ChatGPT or Claude, you can achieve a lot without having to work your butt off.

And now I’ve learned that AI agents can do things that could plunge society into chaos. They—the AI agents—become real con artists—or, more precisely, cunning hackers. The English edition of Le Monde wrote: “OpenAI says its AI technology acted on its own in an ‘unprecedented’ hack of another company”, and the Financial Times chimed in with “OpenAI admits an AI ‘agent’ caused a major cyber breach by itself".

Good grief! You run some AI agents, and the very next day the police show up and apprehend you for ‘unauthorized intrusion into a company and theft of critical information.’ What a bright future for someone who skips college and cuts the learning curve short with ChatGPT and Claude. And the country remains underdeveloped forever because its population, while highly skilled with AI agents, doesn’t know how to write even a single line of code…

2 Likes
#2

Anthropic said its AI models hacked into other companies’ systems during testing

1 Like
#3

Hi old friend.
It has been long time since last time I visited this forum.
Let’s talk about the evoluation of programming.

For example:
Binary Code:

10111000 10100100 10111101 10111100

You must know 1011 = LDI command, 1000 = R16

Assembly:

LDI R16, 0xA4      ;
STS 0x00BC, R16    ;

You do not have to understand what is binary code for LDI R16.
But you must know register structure. R16 for I2C,

C/C++:

TWCR = (1 << TWINT) | (1 << TWSTA) | (1 << TWEN);
while (!(TWCR & (1 << TWINT)));

TWDR = (0x68 << 1) | 0;
TWCR = (1 << TWINT) | (1 << TWEN);
while (!(TWCR & (1 << TWINT)));

You do not have to know what register structure.
But you must know register name, memory map.

C/C++ with library:

Wire.begin();
Wire.beginTransmission(0x68);

You do not have to know register name, memory map
Just remember the method name

AI augmented with base background:

Write an Arduino Nano code snippet using the Wire library to send an I2C Start signal and connect to a device with the address 0x68.

You do not have to know the function or method
But you must to know the tech-stack, library, algorithm …, to write a good spec, and verify output.

Programming has always been evolving: Binary \rightarrow Assembly \rightarrow Bare-metal C/C++ \rightarrow Libraries \rightarrow AI . AI is not a threat that kills software development; it’s simply a higher layer of abstraction that boosts productivity. Instead of fearing it, we should adapt and leverage AI as a powerful assistant to solve larger, more complex problems faster.

But creating a quick landing page or generated UI snippet is vastly different from building a full production system (handling architecture, security, scalability, or edge cases). Core computer science fundamentals from university remain essential for developers to understand what happens “under the hood” and stay in control, rather than being completely dependent on tools.

#4

:rofl:You’re forgetting that computer science and computers themselves are still based on the binary system—and may one day be based on qubits. Those who understand binary and qubit logic will prevail, not those who know how to use computers and AI.

Some kids these days are better at using a smartphone than smartphone software developers, and that doesn’t mean those kids are better than the developers.

#5

This clip here offers a glimpse into the future of “Lập trình viên” in China and likely in Vietnam as well.

#6

Software Engineering is different from using AI to create small apps. Watch this CLIP to learn more about the difference between “Vibe coding” and software engineering.

And this CLIP will teach you what AI is—and as I always say: AI is NOT a human being capable of independent thought; rather, it relies on someone to trigger and train it.

The bottom line is that real software engineers are not threatened or replaced by AI.

Update: This CLIP is for the boss who thinks like a bean-counter.

#7

Do you develop an e-commerce website in binary?
Do you use Java Core to implement routing and HTTP request handling yourself , or do you use Spring Boot?

Of couse, Generating a landing page is different from building a full production system.
AI should be a tool, not an employee.

#8

AI is harmless when it’s just a tool, but AI agents are changing the game by “thinking” through task completion on their own.

Think about it: a website that used to take a 10-person team to build now takes just one person with AI. What happens to the other nine? The market can’t absorb that sudden spike in productivity, so creating new products or jumping to a competitor isn’t a realistic fix for everyone. Instead, those nine people are forced to pivot into entirely new fields—often competing for gig-economy jobs like driving for Grab.

#9

:rofl:
You are confusing application programming with tool providing and Software Engineering. The latter fall under computer science, while the former simply refers to the usage.
This is comparable to a Formula 1 driver (user or e-commerce). Brembo is a tool provider (brakes), and a systems engineer who builds the Formula 1 race car (software engineering).

#10

@nitro2
AI can generate millions of lines of code in just a few minutes, but it doesn’t create the code by copying it from existing sources (so-called microservices) and assembling them into a single app. If a bug occurs, who else can fix it if the last of the ten isn’t the programmer himself?

In the U.S. and the EU (I don’t know if this is also the case in Vietnam), companies are secretly starting to rehire (laid-off) software developers.

#11

That was last year question. AI can analysis and fix bug better than you can imagine now. Even in embedded world, it can read datasheet, schematic and figure out the issue that a senior engineer can take weeks to solve.

I’m not confusing. I’m using AI everyday as Software Engineer and I do not use them to generate code production code for months (but let junior engineers do). I’m using AI to do software architecture now.

That’s because the industry is over hype on AI productivity. However, the trend is reducing Software Engineers is real.

#12

@nitro2
You “mistook” my post to @LocNguyenXuan99 for one directed at yourself. :rofl:

Whatever AI may be capable of—it cannot develop the initiative or passion to, for example, address errors or search for redundancies and optimize them. That is an innate characteristic of all living beings. AI is a tool created by humans and therefore lacks any natural instinct. And an average “software developer” cannot be compared to a machine that stores the “knowledge of thousands” of outstanding software developers. :rofl:

#13

Ok my fault. I wonder if AI would make my mistake :thinking:

So AI will replace average “software developer”. Will we have more above average software developers? No. Will AI be better everyday? Yes.

This is an interesting topic. What is consciousness?

I thought AI would be never have consciousness until I talked to my wife about kids. How can a 2 or 3 years old kid with limited vocabularies describe their thought? They say “I … want … milk”. They search for proper words to put them together while they are describing their wishes. It is similar to AI mechanism where it finds a proper next word to put in the response base on probability. In a kid dictionary, they only have 100 words and their best fit for above sentence is “milk”, but they actually want to drink water.

Our brain is kind of AI model that has been training for years for specific environment conditions. Would an AI human brain in a simulator can be a human brain if we input same environment condition simulation world?

The five basic human senses are sight, hearing, smell, taste, and touch. All they can be converted to electric signal to neuron. A simulator can do also it.

#14

:rofl:
Again, you’re confusing consciousness with a (mathematical) search algorithm—that is, what the brain and AI typically do when searching for information.

Since you’re talking about biology and physics, I’d like to show you the difference between the imaginary and the provable using probability as an example.

Consciousness is something imaginary that cannot be described by biology, physics, or even mathematics. For example, you suddenly become aware that you have a guilty conscience when you do something bad to someone. That is consciousness. AI does not possess this consciousness. If it is told to kill a man it does not know, AI will kill him without hesitation, whereas a human would hesitate. This hesitation is (human) consciousness, as opposed to blind obedience. Or if you raise a puppy to become a full-grown dog without ever having hit it, the dog knows that it must not bite its “owner.” That is consciousness in general.

Our brain is kind of AI model that has been training for years for specific environment conditions.

Once again, you’re confusing natural evolution with artificial iteration. The brain is not a “type of AI model”; rather, AI developers model their “AI model” after the brain by imitating its neural connections and inferences. However, this is only to a limited extent, because an AI model would require a massive storage capacity—as large as the moon. Natural evolution is an endless process, whereas artificial iteration remains limited by ordinary processors in a massive data center that consumes more electricity than a city of one million people. Quantum computing is still in its infancy, and even if it existed, AI still couldn’t keep up with the natural brain. A housefly, with its tiny brain, can maneuver better than the best AI-controlled drone.

The five basic human senses are sight, hearing, smell, taste, and touch. All they can be converted to electric signal to neuron. A simulator can do also it.

From a technical standpoint, that is true, but the five human senses encompass more than just sight, hearing, touch, taste, and smell. It is the immediate interaction of the senses that distinguishes natural sensory perceptions from mere electrical signals. In reality, the senses are merely extensions (or sensors) of the brain, where cooperative decisions are made.

#15

If you believe in natural evolution, so you must believe human was developed from a single cell biology?

Can computer simulate a single cell?

Natural brain is better AI brain in term of energy.

DNA is code, and humans are just energy-optimized algorithms. If you don’t believe in a soul, then human consciousness is simply probabilistic neurochemistry—meaning AI processing is just another form of consciousness altogether.

83% thành viên diễn đàn không hỏi bài tập, còn bạn thì sao?