i agree, i do like the article content itself, but the AI-generated images (clearly nano banana btw) really kill the credibility. even just using stock images with the watermarks clearly visible would be better
it would be great if this were the case. unfortunately, Uber has been documented to practice individual price discrimination at a massive scale, using factors like if you’re in a low-income vs high-income neighborhood, individual rider “price sensitivity”, etc, in addition to market conditions (surge pricing), and as a result they have netted billions in profit [1]. i would guess this is why Uber AI researchers are paid so much.
ah, i see, thanks for the insight! do you have any advice on how one might get started with IC reverse-engineering? i think it would be interesting to reimplement these chips in a way that's at least inspired by the original design
personally, the evolution of replit makes me sad. i remember writing some of my very first ever programs in python using replit in middle school, as it wasn't blocked by the school network and it was the best way of running arbitrary code online back then. i used it to execute java code for AP computer science in high school, and i improved a ton at using the terminal as well. At some point, I stopped using the replit web editor and was coding by full-screening the built-in terminal and using vim. it was a formative experience and really helped me develop as a programmer even though all i had access to was a locked-down chromebook. but now, going back to the website and seeing the first thing it shows to you is how you can "build apps using AI", not even being able to even create an environment to run some python code without talking to an LLM, and the company focusing on ARR and becoming "AI-native" and creating value and all that jazz, and it feels like the magic of learning to code for the first time has been lost. luckily, kids these days are spoiled with webassembly and can run pretty much whatever they want in the browser, so i'm sure the next generation of young programmers will be alright
> Did he consider that the LLM wouldve done it the right way after the first message if prompted correctly?
I think the article is implicitly saying that an LLM that's skilled enough to write good code should have done it "the right way" without extra prompting. If LLMs can't write good code without human architects guiding it, then I doubt we'll ever reach the "10x productivity" claims of LLM proponents.
I've also fell into the same trap of the author in assuming that because an LLM works well when guided to do some specific task, that it will also do well writing a whole system from scratch or doing some large reorganization of a codebase. It never goes well, and I end up wasting hours arguing with an LLM instead of actually thinking about a good solution and then implementing it.