“Hey, no worries - happy to help 🙂
Because of how models work and what they can offload from us (developers), we will be heading to the place where usual repetitive tasks will be completely replaced by the AI.
What cannot be replaced is the architecture and the long-term vision for the project as models are only recreators - not inventors. If I were you today, I'd focus on building scalable and reliable foundations that can be easily interpreted or understood by the AI. This is inevitable, now our jobs will be more like a 10x leverage architects building and guardraling tons of the autonomous agents working like the current junior/mid devs.
I'd for sure learn the very fundamentals of the technologies (like how programming languages work in very very detail on the very fundamental level) at the same time learning how we can connect multiple systems to work like a single body - reducing operational costs, mitigating potential risks with rapid development etc.
Another area might be actually learning and utilizing how agents work and turning their advantages to you own benefits - for example solving real-life small issues (with that you'd learn the business edge - something that AI will never do properly).
There's also another aspect that's nice to have nowadays.... It's ability to troubleshoot, crisp eye and ability to spot issues before they reach production - this is now the most important skill as shortly you'll have a lot of changes made by agents, so you need to learn how to quickly analyze the side-effects and tradeoffs of the given solutions. This also relates to the first point I wrote.
In my opinion we will reach the moment where computational cost of running the models (frontier) would be too high for the most of the people so we will go back to the software engineering (or in some similar way). Then we will see what's gonna be next.
For now focus on the deep level understanding of one or two of the languages you'd like. Utilize AI to help you 😁”
Post #1014
156