Mind the gap

I’ve been into computers ever since I was a child. I took apart my first PC at 12 years old and, surprisingly, it still worked once I’d screwed it all back together.

It was an old server my dad had sourced from work, built like a brick and designed to last an eternity — it had served its purpose and was handed down to me, and I was happy to have a PC of my own and give it a new home.

Like that tired server, I’ve seen many other technologies come and go, frameworks and designs fall in and out of fashion, the slow grind of progress over the years. My generation grew up with analogue and saw the gradual shift to digital — out with physical media and analogue transmissions and over to streaming and on-demand services. If I’m honest I think we’ve gone a little too far, and I miss listening to tapes and flicking through a limited number of channels — more choice is overwhelming and life was simpler back then.

# The greatest achievement

But it’s not all bad — I love technology — and over the past few years we’ve seen the rise of what is, in my humble opinion, the greatest technological achievement of all time. Putting a man on the moon was humanity’s ultimate achievement, but artificial intelligence has the capacity to overshadow anything that has come before. It’s odd, therefore, that it gets written off so easily by many as “AI slop”. I think it’s a disservice to such a monumental and pivotal moment in history.

I can understand the criticism — people don’t want AI rammed into every corner of their lives, into every nook and cranny, into every app — but I don’t understand why more people aren’t excited. AI isn’t perfect — it uses a lot of resources, there are copyright concerns about the material ingested to produce the models, and it can be wrong — but these are challenges to overcome and not reasons to write off the technology as a whole. Much like the first computers that filled a room, I have no doubt that, in time, we’ll be able to scale AI down and utilise it in ways we can’t even imagine today. And much like computing as a whole, it may work its way into every facet of our lives, so much so we might wonder how we’d ever cope without it.

I honestly think AI has the capacity to be the most transformative innovation of our time — greater than the invention of the wheel, greater than the industrial revolution — and I believe we have yet to realise its true potential. Even just a few years ago I wouldn’t have thought we’d see the current level of progress in my lifetime — it truly feels like we’re living in the future.

I think one of the reasons people aren’t as impressed is because they haven’t spent as much time working with it as I have. Perhaps they’re using free models, perhaps they only have a cursory experience with it, but I’ve been using frontier models regularly for some time now and I am constantly blown away by them. I know what AI is, and more importantly what it isn’t, but it can still make me chuckle at times — I’m interacting with an artificial intelligence that is able to make me smile and laugh — how amazing is that?

# In the open

I think it’s only fair to be transparent about where and how I use AI, and I think that’s something we should pivot towards as a society moving forward — we’re overwhelmed with content these days and it’s only fair to denote what’s real, and what isn’t. I use AI to develop code, but the blog, the photos and everything else is my own work.

I justify using AI to develop because I’m already a reasonably experienced developer — I’ve been at it for over 20 years — and so I have a fairly solid grounding myself. I don’t know whether using AI needs justification, but I understand the code it produces and I can work with AI to develop solid, robust applications.

AI has let me transition from bricklayer to architect — and whilst I used to do both, it now gives me more time to think about the bigger picture.

My horizons have broadened and I’m able to create amazing things in a shorter time.

# With great power

But with great power comes great responsibility, and I implore those with less experience to exercise caution. AI lets people produce apps they’ve always dreamed of, but it isn’t a shortcut for learning — without knowledge and expertise AI will just as easily lead them awry, and amazing as it is, with a limited context window, the less attention it can give to bigger projects. It’s up to you, the project manager, to keep things on track and ensure the joins are seamless, else you may find critical bugs slipping through.

But assuming proper care and attention has been taken, then I fail to see why an app developed with AI is worth any less than one without. There’s a famous saying that most people don’t care how something was built provided it works and it’s fast, and I think the same holds true here — no one cares what framework was used, or how elegant the code is; providing your app works, and it’s fast, and doesn’t have critical bugs, then I think it deserves to be taken just as seriously. Ultimately, I think AI is a force multiplier, and in the right hands it can produce awe-inspiring works, but in the wrong hands it can be equally lacklustre and maybe damaging — a little learning is a dangerous thing.

# The storm is coming

For those of you still on the fence, consider this — Anthropic were unwilling to release their latest frontier model because it was too powerful. They instead created a new platform called Glasswing to provide people limited access in order to utilise the new model to harden their code — a responsible approach to stem the inevitable storm. Anthropic’s models are closed source and they’re able to pick and choose who has access, but there are plenty of open source models playing catch-up, and it’s predicted these may be equally powerful within the next 6–9 months, at which time anyone could download and run them at will. So however you feel about AI, it’s worth considering whether it might prove useful for hardening your code, as the storm is coming one way or the other, and I’d rather know the tools I’m using to defend against it are equally adept as those that nefarious actors are using to attack.

In some respects, I think the old days of software development are over. I’m not sure whether that’s a good thing or not — are we ushering in a new technology without thinking about what we’re giving up? Might we find ourselves looking back on the old days with fond memories, realising we’re past the point of no return?

I still think we’re on the right path, but we need to be cautious. We need to ensure technology doesn’t escape our understanding, that it doesn’t get ahead of us, and that we can stay in the loop — rather more worryingly, we need to come to terms with how knowledge and expertise become less restricted. There are plenty of bad actors out there and AI is a powerful tool in the wrong hands, but we can’t keep it wrapped up for long. I think this is the challenge many leaders in the AI world face at the moment, and I hope they find an answer some time soon.

For me, I’m going to carry on coding and enjoy it whilst it lasts — but don’t become too dependent on it.

If you’re nothing without the suit, then you shouldn’t have it.Tony Stark