It’s been over a year I’ve been using Cursor after switching from Github Copilot which at the pre-GPT time did feel absolutely amazing. Let’s reflect back on the past year, the heavy conversations we as developers were having with our colleagues and with ourselves, and what next year can bring.
1.2B tokens later
I’m thinking should I be amazed? Is it a good thing? Is this a bit scary? Probably yes to all.
One thing that this year has been is full of emotions for me as software engineer. Thinking how LLMs are getting amazing with release of Claude 3.5 which to me was the very first model where I could be more hands off keyboard and get something semi decent delivered, to seeing Composer come to life, and Claude 4.5 later this year which both at it’s core seem to be capable of replacing many junior engineers.
We’ve seen many executives embrace AI with open arms, we’ve seen many engineers reject it. And if you are like me with a job in tech, or in one of the affected industries, I bet you lived through this yourself, both the excitement and fear, and I’m almost certain you have your own very opinion that you might even be afraid to share.
3.3k messages & over 600 chats
How does it feel a year later? As the models are getting significantly better, did your fear increase? Do you feel empowered? Are you concerned for the future of vibe coded web?
Myself, I was mainly concerned, especially when you hear as a developer things like “lovable” or “our PM just built this tool for us” – while this is partly a satire, it’s things you hear in reality, and they are partly about empowerment. Giving tools to less technical people that can help them better shape products, with hands on iteration is amazing idea. I remember from my project management job, learning PowerShell scripting, and Python to build tiny automations within my means (extremely restricted computer without install permissions) – was it fun? Hell yea! Was it effective? Yes, at the time.
One thing I remember however, the moment I was able to empower other people with technology that’s exactly what I pushed for heavily. “Here is this new cool tool that’s low code / no code – come learn it, I’ll show you around” – this was on my weekly schedule, I always wanted to push people into new technology because I knew, once adopted it will make them do better as individuals, it will help the company.
148 Sundays?
So what actually is the scary part with AI? Why do so many of developers reject it? Why do we resent the idea of non technical people using it?
It’s really the lack of guardrails in my opinion. I remember one of the first business tools that I loved – it was quickbase. It gave people option to move away from spreadsheets, build automation using Pipelines that ran on schedule, build interactive dashboards, all in one place – yet it kept them safe within their own sandbox.
If you look back at 2020 / 2021 we did see this with DeFi, something new, fresh and exciting. Brand new technology to many, with large adoption, however very quickly also became a significant security concerns, people weren’t used to, and many never got used to the security implications of crypto investing on decentralized exchanges, and many lost significant amounts of money to obvious scams.
This is how I see it with AI today, the introduction of new technology so quickly, while just like DeFi opened up previously unseen opportunities for many, in similar way it failed to prepare the audience to “stay safe with AI”.
I don’t believe after seeing this play out for over a year many developers are concerned for their jobs from the perspective of everyone being able to build great software, I feel like their fear comes from the lack of knowledge of people around them.
8 PM
So what do we do?
Let’s not become the nay-sayers. Let’s keep sharing our passion for technology and let’s embrace AI, but let’s do it with guardrails.
The developer community if I could highlight one thing I absolutely love is how we manage to make learning accessible, let’s keep doing it with AI. While prompting for me feels like is hard to teach and rather is acquired through experience, let’s work on content that promotes online security for non-developers building applications. Let’s educate on basics of software engineering, UX, system design, deployment and many other challenges where you know better.
Let’s focus on the problems of AI era and solve new problems that come with it – Google’s SynthID for authenticity verification, deadend.ai for hallucinated links and whatever you come up with next!
Use AI yourself to augment your workflows, use it to build software that will change the world. All those crazy ideas you had in the past you couldn’t find enough time to build? Your time just got much cheaper. Want to learn something new? Google’s Gemini got your back as your personal teaching assistant. You suck at web design? Your friendly model can do it for you.
None of this replaces actual experts in the industry, but gives you hell of a chance to make something that makes a difference for many people, and that’s what software is about to me.
Note: This isn’t AI rant, or bullish article on AI, just honest reflection of how this year I was feeling about AI from being scared to embracing it and augmenting my own work, launching my own startup – these are also purely my opinions and mine alone and don’t reflect opinions of my employers.
(and yes, the thumbnail is purposely AI generated!)
Top comments (0)