I was wrong about AI

Posted on: 11 June 2026

A little over 2 months ago I made a rather doom-mongering post on my blog titled “AI scares me”. This wasn’t hyperbole; I genuinely felt overwhelmed and a little defensive about the speed and extent that AI has proliferated in almost every industry.

I touched on the fact AI skips the important discovery cycle. I was of the opinion that you’re no longer required to learn how to do something, you can just instruct AI to do it for you. But having used AI, Claude in particular, in my workflow over the last 2 months, I have changed my view.

Learning is still necessary, or at least, it’s still a worthwhile endeavour. AI can replace this, but it can also hugely enhance it. AI is the ultimate sidekick. What was once a lot of head-scratching, fruitless Google searches, and combing through 10 year old blog posts on the off-chance they solved your problem, can now be delegated to AI.

AI has the entire Internet of knowledge in which to formulate the exact answer to the problem you’re trying to solve. And it does it with incredible accuracy and consistency. It still blows my mind just how articulate and specific it can be in its advice to resolving your niche problem.

What’s more, you can use AI to inspect and reason about your codebase. This has been a game-changer for me. You can ask it questions about how something works, challenge it to improve systems, or apply its knowledge of the app to build something completely new.

And as well as being an incredibly knowledgable sidekick, like having a senior developer by your side, it’s also incredibly effective at doing the job of a junior developer. The monotonous tasks that you know how to do, but feel like a waste of time. Like scaffolding some tests. Of course I will verify the output and perhaps add my own touch, but getting me 90% of the way there is quite the time and energy saver.

I still take issue with the usage of AI I highlighted in my original post. You can use it too much; blindly trusting its output, and end up in spaghetti code hell. But as a quick prototyping tool, or used as a mindful coding partner, it is groundbreaking.

However

It’s not all sunshine and rainbows. I still have strong ethical dilemmas about its environmental footprint and the security concerns around it. It’s not yet a sustainable tool, and - in the wrong hands - can easily become costly and unwieldy.

I also shy away from the assertion that AI can be applied to anything and everything. That there are free efficiency gains to be had anywhere and everywhere. Like any tool, it has its purpose, but it shouldn’t be shoehorned in where it’s not effective, just because it’s the new hot thing.

For logical things like coding, I believe it offers a huge leg-up compared to traditional methods. But for creative pursuits, I’m still of the opinion AI is not aiding us. The sanctity of human expression and intellectual and creative property must be protected. The fact there are AI-powered Instagram accounts that have learned from the work of genuine artists and produced “unique” art in their style is a concerning trend.

If we’re using AI to transform bulleted prompts into formatted reports and emails, but simultaneously using AI to summarise these texts back to their key points, what are we accomplishing? Not to mention the incredibly wasteful use of energy required.

In conclusion

Like any technology, we need to guardrail its usage and carefully govern its development. We’re a long way from a dystopian self-aware singularity, but there are a select few responsible for ensuring the safety of humanity.

Yes AI is fun, yes it can make us more productive. But to what end? Are we advancing the human race, or just lining the pockets of the already wealthy stakeholders? Crucially, will it make us less intelligent?

One thing I hope AI does over time is eliminate the so-called “Bullshit Jobs”. Obviously this alone would be catastrophic. But paired with societal changes like universal basic income, AI could help us to make steps towards a utopian future where mundane work is not at the centre of our existence.