Silvestriblog
Aug 212 min read

At last, generative AI is really starting to help

Counting calories with GPT, coding React with Claude, and turbocharging productivity with Cursor. Is this the practical AI revolution we've been waiting for?

It took some time, but I am happy that generative AI is finally starting to shine for stuff other than stealing work from copywriters or make everyone hate you because the email you sent to your colleagues was clearly not written by you.

I remember when just a year ago I'd use ChatGPT for things like that, or to help me with coding - the latter being often an excruciating experience, with the main limitation of a cut-off date that couldn't keep up with the evolution of frameworks and libraries. So eventually, I let it go for a few months, and used it very sporadically.

Fast forward to today, and I'm using ChatGPT to help me calculate the calories of the food that I eat when I'm out (it's not 100% accurate, but it's not so far off from reality either). I use Perplexity instead of Google when I'm looking for specific answers, and most importantly, I employ Claude 3.5 Sonnet when it comes to React development. And it's making a huge difference.

I didn't expect that Antrophic would come out with something that is just so much better then GPT at coding, but I'm glad we're here. Now we just need to sort out that naughty little habit of smacking servers around the globe millions of times in just a few hours...

But what really has surprised me is Cursor. This is a modified version of Visual Studio Code that heavily relies on AI to provide a number of functions that are really speeding up the development process. We are talking about a "hey, AI, literally write the app for me" degree of automation here.

I discovered it on X but initially hesitated due to its lack of GitHub Codespaces support. But yesterday, I finally decided to ditch Codespaces in favour of local development, and the results were astonishing: I accomplished more in the span of 3 hours than I had in 3 days without it. I regret not adopting it sooner.

The evolution of generative AI is now starting to feel a little more positive, at least in my industry and from my perspective. As development becomes more accessible, we'll have more room for innovative ideas and less time wasted on technical minutiae.

It can't be that bad, right?