The future is here. Apparently. Or are we just clutching at straws, hoping for the best, because, let’s be honest, these AI tools aren’t (yet) as good as they claim to be, are they?
Oh, they promise the world. Revolutionising industries, making our lives easier, doing things “better” than we ever could. And yet here I am, questioning why I need a fistful of cocodamol just to deal with them.
Yes, I have a pounding headache!
Cursor: The Visual Studio Code Fork That’s Sometimes Useful, Sometimes a Bloody Liability
Let’s talk about Cursor. If you’ve never heard of it, well, lucky you. It’s basically a fork of VS Code with built-in AI capabilities, claiming to be the coding assistant of the future. Essentially, it’s VS Code but with extra AI magic sprinkled on top. At least, that’s the idea.
Now, I like testing tools. I like seeing what works, what’s a waste of time, and what makes me want to headbutt my desk in frustration. Cursor falls somewhere in between.
Because sometimes, it works. Sometimes, you throw a problem at it, and it gives you an actual solution. A rare but pleasant surprise, much like finding a McDonald’s ice cream machine that actually works. But other times? Other times, it fails spectacularly, and the only solution it has to offer is: turn it off and turn it back on again.
Seriously. This is where we’re at in 2025.
Control | Alt | Delete.
The ‘Unable to Reach Anthropic’ Error – A Classic IT Blame Game
Cursor integrates with Claude, the AI model from Anthropic, but there’s a fun little game it likes to play: the ‘Unable to Reach Anthropic’ error. And this, my friends, is where things get properly irritating.
You see, when this error appears, you’d assume Claude’s servers are down, right? Wrong. Claude is absolutely fine. Up and running. You can test it in other ways, and it works perfectly. The problem is Cursor. But does Cursor admit this? Of course not. Instead, it just sits there, arms folded, refusing to do anything useful, like a sulking child who won’t explain why they’re having a meltdown.
It’s classic IT bullshit. Like when your website won’t process payments, and the payment gateway blames the website. Then the website blames the gateway. Then the gateway blames the bank. And the bank blames the customer’s card.
Nobody takes responsibility.
That’s exactly what Cursor is doing. But here’s the kicker (wait, no, I don’t say that, let’s just get to the point) – if I’m paying for a tool, I don’t want to be dealing with “capacity limitations” and “connectivity issues.” This isn’t a freebie. I’m not here for a trial run of “AI roulette” where I might get a working response or might get told to piss off and try again later.
The Genius Fix? Shut It Down and Restart. Seriously?
So, what’s the grand solution to all of this?
Turn it off and turn it back on again.
Yep. That’s the advice. Restart the session. Restart the software. Start a new chat. Maybe even restart your computer just for fun. Because, apparently, this is the height of AI troubleshooting in 2025.
We’ve got multi-billion-dollar AI models running complex operations, but when they fail? Oh well, just reboot it like it’s Windows 95 having a stroke.
What if I’m halfway through a coding session and need continuity? What if I’ve spent ages refining prompts and debugging? Nope. Tough luck. Start over.
Imagine if this logic applied to everything in life:
- Doctor: “Ah, you’ve got a headache? Have you tried switching yourself off and on again?”
- Mechanic: “Your car won’t start? Have you considered turning it off and back on?”
- Pilot: “Engine failure? No worries, we’ll just reboot mid-flight.”
This is why I won’t drive an AI-controlled car. Because I don’t fancy rebooting my vehicle at 70mph and hoping it remembers which way I was going.
Is This Really the Future? Or Just a Fancy Version of the Same Old Crap?
We keep getting told AI is the future. That it’s going to change everything. But if the best advice they can give when things go wrong is “restart the bloody software,” then I’m not convinced.
This isn’t progress. This is just fancy rebranding of the same half-baked, unreliable tech we’ve been dealing with for decades.
AI is supposed to make things easier. Yet, half the time, we’re bending over backwards trying to work around it. What’s the point?
If AI tools are going to be truly revolutionary, they need to actually work. Consistently. Without needing daily resuscitation.
But until then, we’ll keep doing what we’ve always done: swearing at our screens, restarting things like it’s the bloody 90s, and questioning why we ever thought the future was going to be any better.
AI, Cursor, and General Tech Bullshit – Burning Questions Answered
Does Cursor Actually Work, or Is It Just Another AI Gimmick?
It works — sometimes. Other times, it’s about as useful as a chocolate teapot. If you enjoy playing roulette with your coding assistant, go ahead.
Is Cursor Better Than GitHub Copilot?
Not really. It’s like asking whether you’d prefer a left hook or a right hook — different styles, same pain. Neither is flawless, both can frustrate.
Why Does Cursor Keep Saying It Can’t Reach Anthropic?
Because blaming Anthropic sounds better than admitting their own servers are wobblier than a toddler on roller skates. It’s a polite dodge for downtime.
What’s the Best Way to Fix Cursor When It Stops Working?
Close it. Reopen it. If that fails, breathe, question whether you really need AI for this, and maybe try something that won’t raise your blood pressure.
Is AI Ever Going to Work Properly, or Are We Stuck in an Endless Loop of “Turn It Off and On Again”?
Good question. AI keeps evolving, just mostly into shinier versions of the same old nonsense. Keep your Task Manager ready and hope one day it simply works.