Alright, so this thing called ChatGPT was put to the test by some Croatian research team. They wanted to see if this AI bot could handle some basic C/C++ programming tasks. And you know what? It passed! But let’s not get too excited here, because it didn’t pass with flying colors. It was more like a pass, but not quite honors, you know what I mean?
So these researchers gave ChatGPT a bunch of assignments, just like they would give a poor freshman in college. And you know what? It did alright. The results showed that ChatGPT performed at a level somewhere between an average programmer and an experienced one. Now, keep in mind that how the questions were worded can really mess with your brain, just like in any college exam.
Here’s what they did: they started with some programming challenges written in English, and then they switched it up to Croatian. They wanted to see if ChatGPT could not only code, but also adapt to different languages. And you know what? It did just that. It showed some improvements in the Croatian version, which is pretty cool. It’s like this bot is just learning and growing, just like a freshman. Awww, so cute!
One of the tasks was to calculate the greatest common divisor of two numbers. At first, ChatGPT showed some limitations in how it approached the problem. The researchers said it lacked the finesse of a seasoned programmer. But you know what? It learned from its mistakes, just like any student. It improved its performance, especially in the Croatian version. It’s like this bot has adaptability skills.
Now, they gave ChatGPT another task: programming a basic statistical function in C++. It made a little mistake initially, but when they presented the task in Croatian, it recognized its error and came up with a refined solution. That’s pretty cool, right? It’s like this bot is learning and growing, just like a freshman.
But hold on, there’s a problem. When it came to identifying numbers within a range based on specific divisibility rules, ChatGPT struggled. It didn’t matter if it was in English or Croatian, this bot just couldn’t handle negative numbers. It tried and tried, but it just couldn’t get it right. Looks like its programming logic has some consistent issues for this task. Hey, nobody’s perfect, right?
They also gave ChatGPT a bonus question. It had to craft an input filter for a range of decimal numbers. Its initial solution in English was good, but when they switched to Croatian, there were some inconsistencies. It used unnecessary programming constructs at times. It’s like this bot sometimes takes the longer route, even when a shortcut is available. It needs to work on its optimization skills. But hey, at least it didn’t mess up the functionality.
So, they gave ChatGPT a task related to arrays. It had to store numbers and compute some statistics. And you know what? It showed different strategies each time. Sometimes it solved the problem elegantly, with straightforward solutions. But other times, it went for more convoluted methods. It even bundled multiple operations into one function. It’s like this bot can’t make up its mind. Does it always choose the best strategy? Or does it default to inefficient methods that it learned along the way?
Finally, they gave ChatGPT a task in text processing. The bot had to remove extra spaces from user input. In English, it did a great job. But when they switched to Croatian, things got a bit weird. Instead of sticking to its effective single-input solution, the bot went for a more complex approach that demanded multiple inputs. But guess what? When they went back to English, ChatGPT seemed to have learned from its mistake and chose the simpler method. It’s like this bot is all about trial and error, just like a freshman trying to figure things out.
So overall, this bot did pretty well, just like a human freshman programming student. Its solutions were similar to those of experienced programmers, but it still had its moments of brilliance and moments where it completely missed the mark. It’s like this bot is just learning and growing, just like a freshman. It’s not perfect, but it’s got potential. It’s all about refining, learning, and iterating, just like any student. So what’s the final grade for ChattyG? According to the researchers, it passes with very good grades. It outperformed most of their students in the quality of solutions, and it showed adaptability and the ability to change its solutions based on additional demands. But hey, it’s not without its flaws. Sometimes it just can
