00:00:00At this year's World Economic Forum in Davos, Dario Amodei predicted that within the next
00:00:0712 months or so, AI would be able to write all code fully automated on its own, essentially.
00:00:15And it's worth paying attention to what this man has to say. And I'll also share my thoughts on it
00:00:20and why I think that you should take a more nuanced look at that. Not just because Dario, of course,
00:00:26is the CEO of Anthropic, which is one of the most important players in the generative AI space,
00:00:32especially when it comes to models related to coding. But it's also worth paying attention,
00:00:36because last year, in the beginning of 2025, he predicted that AI would be able to write
00:00:4490% of all code within the next three or six months. And depending on how you look at that,
00:00:50he was not entirely wrong. Now, sure, this timeline is probably or was probably a bit too aggressive.
00:00:58It maybe wasn't three or six months, more like six to nine months, not sure. And of course,
00:01:05it totally depends on your field. So in which area you're programming, also on the programming
00:01:11language you're using, it will depend on your company, the policies you have in your company.
00:01:15And of course, it also depends on you, your preferences, your experiences with AI.
00:01:21I can say that for me, I am at a point I would say since maybe December, November, where for
00:01:29some projects, most of my projects, AI does write 80-90% of all code for me, probably. But this is
00:01:39now where things get interesting. I'm not talking about wipe coding, and I'm not talking about AI
00:01:47doing it all on its own. And I'm also not talking about AI getting it all right. Instead, things are
00:01:54way more difficult than that. So it does absolutely not do that on its own. It also is not about wipe
00:02:05coding here. I mean, you can probably wipe code, but I've shared my view on that before. And for me,
00:02:11by the way, just to make sure we all get the same definition, wipe coding really means that you don't
00:02:16look at the code at all, that you don't care about the code, that you don't necessarily even know how
00:02:21to code. So yeah, this is also not about wipe coding here. Instead, I would say we are at a point
00:02:30where you can hand off many tasks to AI with a good plan. So if you have a good plan, AI can implement
00:02:42it. So I think we're at that point right now. You can definitely do that, at least for some tasks.
00:02:50But of course, you want to review that code. And I know there are people that tell you,
00:02:57you don't need to do that anymore. Might work for them. Definitely not working for me. And
00:03:02if you're shipping anything, if you're shipping anything to customers, you're responsible for
00:03:07that. You as a developer, you can't say, "Oh, the AI got that wrong." No, it's your responsibility.
00:03:13And I won't take responsibility in code I haven't reviewed. I don't understand. Also, still,
00:03:21AI makes many mistakes. So I need to fix mistakes or steer AI in the right direction.
00:03:32And that is also very important. That's of course important to the planning part or related to the
00:03:41planning part that you steer AI in the right direction. But it's also related to what you
00:03:46do with code AI gives you. So very far away from AI generating 90% of the code. And that means
00:03:54I have no work to do. Far away from that. It just means that I got a very fast typer who can
00:04:03implement my plans, but those plans need to be good. And the output is kind of varying in quality.
00:04:11Still, for me, it's very likely faster than if I would write it all from scratch. If I instead
00:04:17try to build good plans, split that up into smaller chunks, let AI write the code for me,
00:04:22and then review and fine tune that code. And by the way, with planning, I mean really detailed plans,
00:04:28where I also break down the exact libraries I want to use, the patterns I want to use,
00:04:33the architecture of the software I want to implement step by step. So it's not a rough plan
00:04:38or a general plan. It's a very detailed plan. That works for me. And therefore, for me, with
00:04:44these restrictions here in mind, I would say, yeah, it can probably write 90% ish of the code.
00:04:52But that does not mean that it does 90% of the work on its own. Now, that's important context,
00:04:58because that's referring to last year's statement by Dario. So yeah, I would say we are kind of there,
00:05:05but not in the sense of AI doing it all on its own. Now, what about this year's statement? Though in
00:05:12this year, Dario essentially said that AI will be able to do what software engineers do and write
00:05:17software fully on its own end to end within six to 12 months. That's this year's statement. Now,
00:05:24I fully recommend watching this entire talk. It's very interesting. But I have some thoughts about
00:05:30this statement. And obviously, just to make that very clear and obvious, I'm in no way smarter or
00:05:36more capable in judging the performance of AI models than Dario. But I'm also not the CEO of
00:05:43a company that needs to sell these AI models. And I can share from my own experiences. Now, as I said,
00:05:50I would agree with last year's statement to some degree, but with many caveats, with many
00:05:56restrictions, the AI is definitely not writing 90% of my code on its own. So naturally, I have a very
00:06:03hard time imagining that this will come true. And I have a very hard time imagining that this will
00:06:08come true, not just within the next six or 12 months, but anywhere in the near future. Now,
00:06:15I totally see that AI is capable of building software on its own in a loop with the Ralph
00:06:24loop that's getting a lot of hype related to cloud code. I totally see that. But the full work of a
00:06:33software engineer includes the tasks I outlined here. And arguably more than that. It includes
00:06:40building a good plan, defining which architecture, which patterns, which technologies to use,
00:06:46reviewing the code. And also, of course, analyzing the code, fixing problems with the code,
00:06:54taking responsibility for that code. That is something I have a very hard time seeing
00:06:59within the near future, because right now, the AI I can use is too far away from that. It's a talented,
00:07:08fast writer that makes a lot of mistakes along the way, and that needs very clear guidance.
00:07:14And that going to models that can do it all on its own, that can plan out entire architectures
00:07:20cleanly on their own, that are capable of using the latest technologies, of writing error-free secure
00:07:28code totally on their own without review or only with review by other AIs. That is something I have
00:07:36a very hard time seeing when I look at today's models and also when I look at the progress of
00:07:42models over the last years. Because yeah, that progress, of course, has been remarkable. It has
00:07:48been steady and good. And especially the tools, I shared that before, got way better. So I shared
00:07:55it before with models. I'm not sure if for the raw model intelligence, we're still on a linear
00:08:00leave-alone exponential trajectory. For the tools, I would say we definitely are on some linear
00:08:07trajectory. But I have a hard time seeing that being enough to get us to that full automation
00:08:15anytime soon. Now, naturally, these CEOs have other interests than me. And you could say I have the
00:08:24interest of protecting us developers also because I also sell programming courses. But let's be totally
00:08:31honest, that is not my role and nothing I will be able to do. I'm just sharing here what my experience
00:08:38with AI has been. And I'm very open-minded, I would say, for AI. I'm using it a lot, as I said. It does
00:08:44write 90% of my code. It's just far away from that full automation. But yeah, please let me know your
00:08:51thoughts on this as well. Your experiences with AI and also in which field you are working. Because
00:08:56you might not be using AI at all or only for small isolated tasks. And you might still write most of
00:09:02the code on your own. Or you're using it for everything. You're not even looking at the code
00:09:07and you're having great results. I'm interested to learn more about that. So please let me know.
00:09:12And yeah, have a great time.