A fictional dialogue between a software engineer (swe) and a friend about AI in Tech.

Friend: So what are you currently doing for work?

SWE: I'm a software engineer. My work increasingly involves asking my AI agent to do things and then I push some buttons. It's a pretty cool job.

Friend: On the topic of AI, are you not worried that AI would take your job?

SWE: No, not at all. Software systems are very complex, stakeholder management even more so. There's still a lot of thinking involved that AI just can't do. We're also collectively learning how to use AI well — prompting, knowing when to plan, giving LLMs the right context, understanding different models' strengths and weaknesses, picking the right tools, and spotting when your AI is hallucinating. There's a real art to using AI effectively now.

Friend: Okay, so if not now, what about the future when bigger and better AI models exist? Artificial General Intelligence (AGI) and all that?

SWE: Well, AI — language models specifically — have improved a lot in recent years, but the technology is still the same at its core. It's a next-word prediction model built on the Transformer architecture, just scaled up. I'm not sure what the performance ceiling is, but some research suggests there is one. And that's before you account for data limitations, compute costs, energy, infrastructure, and all the alignment and safety challenges.

Friend: So there's still a lot we don't know about AI. Would you say you can do your job without AI?

SWE: Well, yes, but I wouldn't want to. I've been writing code — though maybe not truly "engineering" software — since before AI went mainstream, and life with AI is just nicer. AI is a powerful tool that helps me get my work done faster and sometimes makes software engineering more enjoyable. Repetitive tasks are easier and debugging is easier. It doesn't always get things right but when it does, which it increasingly does, it saves me a lot of time.

Friend: So you are saying AI is great for you in your job in software engineering but how about the rest of us? What about people who don't know how to code or are not software engineers?

SWE: It's funny you ask that. Ben, a friend of mine who knows nothing about code, is currently building a mobile app — not using no-code tools or those app builders you see in ads everywhere. He's building it the way a multi-national company or VC-funded startup would build a mobile app. AI is helping him design the user interface, create the backend logic, set up databases, and it's writing all the code for the full-fledged app.

Friend: Oh wow, that sounds impressive. So AI is changing the game for non-technical people too?

SWE: Absolutely! It's empowering people like Ben to build software. That said, it hasn't all been smooth sailing — I've helped him with a few things, given him some tips, and I'm currently helping with testing and feedback.

Friend: Hmm, if he's able to do that, is that any different from what you do as a software engineer?

SWE: Yes it is because Software engineering is more than just writing code. The way I like to describe it is that a professional photographer with a smartphone camera can take better photos than an amateur with a high-end professional camera could. The professional photographer understands lighting, composition, and post-processing techniques that make a photo stand out. And if the camera doesn't work as expected, the amateur won't know how to fix it.

Friend: I've really enjoyed our conversation. You've given me a lot to think about regarding AI and its impact on various professions, especially people new to the workforce. Next time, I'd love to dig into the environmental impact and safety concerns too.

SWE: Definitely! Those are important topics to discuss. Looking forward to our next chat!