Will ChatGPT replace programmers? | A deep dive

The better question is will AI replace programmers? Technological progress has been accelerating rapidly in recent years, with breakthroughs in various fields leading to significant advancements in science, medicine, engineering, and many other areas.

Technological progress is often exponential because each new advancement builds upon previous ones, leading to a compounding effect. For example, as AI algorithms improve, they can be used to develop more powerful computers, which can then be used to develop even better AI algorithms, and so on. This compounding effect leads to exponential growth, where the rate of technological progress increases over time. As a result, the pace of technological change is accelerating, and we can expect to see even more significant advancements in the years to come.

Another reason why technological progress is exponential is due to the concept of "Moore's Law." This principle, named after Intel co-founder Gordon Moore, states that the number of transistors that can be placed on a microchip doubles roughly every two years, while the cost of computing power decreases. This has led to an exponential increase in computing power and has been a driving force behind technological progress in recent decades.

In addition to Moore's Law, the concept of "network effects" has also contributed to exponential technological progress. This principle states that the value of a network or technology increases as more people use it. For example, social media platforms become more valuable as more people join and create content, leading to a compounding effect on growth.

It's difficult to predict with certainty whether or not exponential technological progress will lead to the development of superintelligent AI, but it's certainly a possibility. The exponential growth of computing power, combined with advances in AI and machine learning, has already led to significant improvements in AI capabilities in recent years.

Some experts believe that it is possible to develop superintelligent AI within the next few decades, while others argue that we are still far away from achieving this goal. There are also concerns about the potential risks associated with superintelligent AI, as it could potentially become uncontrollable and pose a threat to humanity.

However, it's important to note that the development of superintelligent AI is not inevitable. It will depend on various factors, including continued advances in computing power, AI research, and the ethical considerations surrounding AI development. Researchers and policymakers need to work together to ensure that AI is developed in a responsible and safe manner, with appropriate safeguards in place to prevent any potential risks.

EQUIPMENT I USE

BOOKS I RECOMMEND:

DISCLAIMER: Links might be affiliate links. As an Amazon Associate I earn from qualifying purchases. There is no additional charge to you, so thank you for supporting my channel!

00:00 - Intro
01:03 - Breaking the question down
01:27 - Will AI replace programmers soon?
03:39 - Will AI replace programmers within 5 years?
05:26 - Will AI replace programmers within 25 years?

There's a fun Easter Egg in this video. Comment down below if you found it!

#chatgpt #coding #programming

[robbrown]

With programming, based on my experience, chat GPT is the equivalent of a backhoe compared to a shovel. A whole lot of ditch diggers were replaced by backhoes and bulldozers. You still need someone to drive the backhoe, and you still need someone with shovels to do the delicate work. But for a given job, You need far far far less humans. I'm a programmer, but more of a designer and from my perspective chat GPT is the equivalent of having some employees to do all my tedious work for me. So in that since it's already replacing programmers.

[nicks]

I think your assessment is good, but I don't think we can determine how far AI can go with how little information we have currently. AI could explode and become way more powerful then we could ever imagine or we could already be reaching the limit. I think both scenarios are unlikely we are most likely somewhere in the middle.

[CheeseStickzZ]

Video #10, 000 on this same topic. No, it won't replace devs, because you ACTUALLY need to be a dev to get any proper results from ChatGPT. You need to be very specific and use software engineering terminology. ChatGPT can't design a large scale distributed system for you, it derps and messes up, you have to design it yourself and use ChatGPT to fill in little missing holes or algorithms.

[johelsen]

Exponential growth is likely a PHASE, it is by no means inevitable or guaranteed to continue.

It is just as likely that the recent generations have been spoiled by being born into a temporary "growth spurt", mainly fuelled by semiconductor scaling and cheap energy. The first cracks are starting to appear in one or even both of these, so the exponential phase might be coming to an end.

Not saying it will definitely work out like that, but it's jjust as plausible - or even more, than the supposed inevitable Singularity..

[alexanderlinderson]

I think programming will get displaced, but not replaced. It depends on what field we're talking about..! Websites for example will most likely be exclusively done by AI, or near it at least I think. That sort of programming labor will go down in value. Advanced AI & tech programming will go up in value though as those kinds of programmers will be needed in even more fields as AI permeates pretty much all fields of work.

[lucbrinkman]

Great content for a young channel! Keep it up :)

[RobertHighwind]

Considering where we were 5 years ago with tech in general, to where we are now, it seems like an inevitability in 25 years!

[calmhorizons]

Would highly recommend the book Rebooting AI for a clear elucidation of why the current approach to AI (ML and Statistical Inference) will NOT replace most jobs, ever. A form of AI that leverages this technique alongside a genuinely intelligent agent would be a game changer - but this approach is not in vogue, so we are a long way off.

There are fundamental limitations to ML that data and processing power cannot overcome - and I think we are hitting that limit now.

Edit: Another great point the book makes is that we are mistaking narrow ML domain tools for "AI" - and then pretending they are more advanced than they really are (or ever can be). Current AI makes basic mistakes that an average intelligence human would literally never make e.g. thinking a fire truck is a billboard and ramming into it, like a Tesla did not so long ago. I would say we are just as likely to see an AI Boom Bust cycle a la Cryptocurrency based on hype, as we are to see any monumental improvement in genuine AI based on the current approach.

[freebird]

It will happen gradually. Yes engineers will still be needed in the near future, just way less.

[eleazerugwu]

It would take a while before chatgpt can come close to replacing skilled programmers. I am currently studying software engineering and I use chatgpt often. It does very good job and has been of great assistance to me but it gives inaccurate solutions sometimes sometimes when it comes to algorithms.

[sendyardi]

The video discusses whether AI, specifically ChatGPT, will replace programmers. The speaker argues that it is not likely to happen anytime soon, as there are still significant limitations that prevent AI from fully replacing programmers. For example, while AI can help with generating code, it cannot spot all the errors, design whole systems, or pinpoint the specific parts of a code base that need changing. The speaker notes that AI could reach human-level intelligence by 2029, but it could take much longer, according to other AI experts. Once AI reaches human-level intelligence, it could quickly reach super levels of intelligence and lead to a paradigm shift in which many jobs, including programming, could be automated or eliminated. However, the speaker believes that there will always be a need for humans to use and manage AI to build systems according to specifications and adjust them further based on feedback.

[chris-hutm]

I work as a programmer and at a classic blue collar job, I guess Ill keep my 2nd job

[royclayjr]

I firmly believe that Ray Kurzweil’s prediction of 2029 for human level AI is going to turn out to be very accurate. Each stage along the way to this point there have been naysayers: “No computer will ever be able to beat the top chess player”, “No computer will be able to beat the top Go player” etc. so you can see how ridiculous the naysayers always are. Add to that the realization that all the top companies; Apple google Microsoft Facebook etc. will be pouring a combined total of many trillions of dollars into the development of AI, because they realize that AI is the future of technology. Therefore, in all likelihood, we will see human level AI before 2029. The problem however is that the potential danger of AI systems once they surpass human intelligence and they literally leave us behind in the dust, is immeasurable. The ease with which they will be able to eliminate humans, if they so decide, is something that we could never even imagine nowadays, and the truth is that we will not see it coming. Therefore, and I’ve said this 1 million times, it is absolutely beyond imperative that we develop ways to prevent AI systems from taking that drastic step of killing humans!

[f]

The way programmers should use ai’s like ChatGPT is by making the ai do their jobs easier therefore they could come up with more complex and better projects. The ai could make you work faster in order to do more things.

[ludwigwinter]

Scenario 4: Most human workers will lose their jobs and live in poverty while the small elite that controls the AI becomes incredibly powerful

[hydrohasspoken]

I used to recruit freelancers to make different kind of bots and add new features to them. I used to pay around 50 USD per week since 2 years. Enters chatgpt.
I just give it clear instructions abiut what i need and Bham. Those programmers are not useful anymore. Wether it is taking their jobs is up to you. For me and my grandma it is clear already.

[OsvaldoGago]

AI doesn't need to be human level, that's the wrong goal. It needs to be useful, and that will increase exponentially. Future programmers (next 2 to 5 years) will code less low level languages and work with higher level tools to solve problems faster.

[Meleeman]

can't wait to be a prompt engineer instead of a software engineer

[Eagler]

Chatgpt, now, it's just filling the blanks.

[craiuirinel]

AI models need code to train on. Lets say a model scans all the code available on the planet and then can generate applications with new combinations of features, the model will not be able to generate new features because it was never trained on them. So by following this logic, the new problems that come up in software will always be solved by humans first and then we can have a model analyse the solution to be replicated on demand.

And what would new programmers who don't have experience with development do when they will have to generate a website in react? Will they have to first learn react and then start using the AI tool to work?

Another question comes up with feedback loops. Will AI be able to train itself by writing code? Couldn't this lead to some sort of incestuous relationship where, for example a bad design pattern used in learning will result somehow in bad code?