This is some of the best analysis I've heard. As a person who has built applications for 7 years now, I thank you. Most of the negative comments seem to be from people who have never actually built software.
The author is wrong. The fact that business will always need a lot of software and that is why programmers will always be needed in millions is no argument at all. AI is drastically changing the way we write software: instead of months, spent on planning and making it up, we will use Ai promts when talking to AI how to design or test software. We will tell AI something like: Build me this or that system for air flight control, make the system be based on this or that tech stack, put the Red button here and there and so on. So, instead of dozens and hundreds of programmers' teams we will need just one human software design AI operator. This is how it is going to be soon.
While that may be the future, I don't see any evidence of it arriving "soon". The "build me this" scenarios don't consider the increasing levels of complexity in complex systems. Programmers will become obsolete when thinking humans become obsolete. That could happen with AGI, but the implications are much bigger than programmers. Keep in mind that programming is a creative task.
You're right we're going to need more programmers. The only problem is everyone will be a programmer. Just prompt an AI. Words are just the next layer of abstraction
Yes "everyone" will be a programmer, but there will continue to be stratifications of skillsets and challenges just as there is today. This isn't a problem, just more opportunity.
The author is wrong. The fact that business will always need a lot of software and that is why programmers will always be needed in millions is no argument at all. AI is drastically changing the way we write software: instead of months, spent on planning and making it up, we will use Ai promts when talking to AI how to IN SECONDS design or test software. We will tell AI something like: Build me this or that system for air flight control, make the system be based on this or that tech stack, put the Red button here and there and so on. So, instead of dozens and hundreds of programmers' teams we will need just one human software design AI operator. This is how it is going to be soon.
Historically all innovations that increase programmer efficiency have increased demand for software and programmers. Will that continue? Is AI truly different? I guess we will find out.
@@prolegoinc Innovation? Are you sure that AI refers to "innovation", as you said? I do think that it's not. The wheel, typography, internal combustion engine, electricity, telecommunications, AI ... these are not innovatioins, they are the turning points in human history that have significantly influenced the course of human civilization. And this is why your video is wrong.
This will not age well. I am laughing when I see people talk about how developers will not go away because to make software you need to understand the client and be able to think, assuming that AI (or the AGI that we will have within the next 12-16 months) will never be able to do it. I've got some news for all these people. GPT-4 can already understand anything a human may need through natural language, and it is also better than most humans at it. Just taking a look at what it can already do with AutoGen should be enough to convince anyone with at least some experience, but people just want to live in denial about the current situation. There will be no need for human developers within the next 2-3 years, because anyone will be able to use AGI agents to have them make anything they may need, but they will make it within a few hours if not minutes and for only a couple of dollars. They will also include any other kind of asset required (images, UI elements, sound, etc.) for the software.
Autogen is interesting, but unfortunately it isn't useful: ru-vid.comHPLEqgIzgKw I can see your scenario appearing if we do get AGI - in that case human thinking generally will be replaced. But it won't happen with LLM Agents even if we have another big leap forward like GPT 3 to GPT 4. Orchestrating and directing agents is just way too hard, and GPT-4 is pretty lousy at it.
You're saying this as if lacking basic common sense isn't the biggest problem with current generative ai. For example if you say you have two buckets one that's 12 L and one that's 6 L and ask it how it would measure out 6 L from a water can the AI loses its mind instead of immediately using the obvious solution.
