Hacker News Reader

Comment by rco8786 | Hacker News Reader

This is a similar reaction folks had to autocomplete/intellisense when it came out also. "What you just press tab and get all the possible options. Psh, how about reading the docs like the rest of us did?". I think AI is a pretty big step function up in terms of the computer's capability to predict the code you want - but at the end of the day we'll still be writing code (I think. I hope).

View on HN

Overreliance on intellisense does worsen your abilities, though, especially when dealing with complex packages and APIs. So does stack overflow. In my domain of data science and finance, it is VERY noticeable when someone doesn't actually understand how Pandas and Numpy work, for example, and how to write good vectorized solutions to their problems

It doesn't help that AI generally creates very mediocre Pandas code (based on a lot of the training data showing mediocre practices).

Pandas is a very difficult api to use correctly and a huge swath of programs that the AIs have trained on was "programmed by accident" meaning people just typed stuff at it until they got the output they wanted.

To offset all the bad code, you would have to make a Pandas fine tune and/or ablate the bad Pandas from the weights.

Does anyone have good reference material on learning how to use pandas effectively and understanding the api as a whole?

Any time ive had to use pandas I am shocked at how convuluted and opaque the docs are, and end up just hacking away till something wirks alright.

I felt the same - have to relearn/lookup everything every time I went back to a project or wanted to do some operations that are simple to describe in SQL but I couldn't wrap my mind around e.g. using multi-indexed dataframes & aggregations properly. These days, I always jump to Polars instead of Pandas - much more intuitive and consistent API. Tons of props to Pandas for all that they did (and continue to do) in the data space, but their API did not evolve very well IMO.

I've also been wanting to play with Ibis[1] recently, but Polars has been sufficient for me.

[1] https://ibis-project.org/

There’s Pandas for Everyone by Daniel Chen.

What helped me with Pandas was a (very) short stint with array programming, specifically uiua. It gave me a good understanding of the possible operations on arrays.

Is this a setup??? :)

The guy two comments up wrote a great book on using pandas effectively. It’s called Effective Pandas

;)

You might be right that intellisense is to blame, but my suspicion is that some people are just worse at writing code and it isn't really about the available tools.

With the difference that intelli* is usually correct - i.e. it fetches all the possible methods from actual code, and you can see the related api docs. It’s another view for the same data (more or less - of course there can be docs other than api docs)

With AI you’ve got no idea whether something is right, it’s a probablistic thing

That’s why I don’t buy the “AI as just another abstraction layer” argument. AI is something different and we should use it differently.

I also wonder if people truly learned their IDE. Even xCode with all its warts have nice feature for completion, debugging, profiling, documentation,…

It seems like all they know is VS Code which is neither a good editor, nor a good IDE.

From my experience, people are really wary of learning the tool they're using to code, because any editor similar to VSCode fulfills their needs.

However, as you said, mastering a proper IDE is a superpower and allows much more to be done in a single window in a shorter amount of time.

Also, since VSCode is everywhere, getting people off of it, or showing much capable software is a sad and moot effort, from my experience, again. KDEs humble KATE is much better than VSCode on many fronts.

> Also, since VSCode is everywhere, getting people off of it, or showing much capable software is a sad and moot effort

The issue is that they're not learning VS Code. They install a lot of plugins, but they're not really learning them. I've seen people that still go through the explorer to open a file.

It's not limited to VSCode. Except some hardcore Emacs/Vim circles, I don't see any people trying to understand and use the power of the tools they have at hand.

They even find configuring an editor or an IDE as wasting time, and of no use. Then, when you show how you handle some stuff, they ask "how did you do that??!??".

I think Copilot (and all other AI coding variant) is different from auto-complete. Auto-complete just give you function names, variable names, but AI coding can write whole block or even file of code, if you so wish.

By the same principle, I also think advancement in AI is different from other technological advancement. Even the invention of computer is at most on par with AI if AI is able to walk far. People always try to use the train-horse analogue but I think we will see a gloom future in the next 5-10 years -- especially when the whole world is not turning to the left, but to the right.

Now that US and China and everyone else are competing on AI, that future might come earlier than I thought.

I do read the docs a lot less now though with IntelliSense, so if my craft were reading and understanding docs I'd probably be measurably disappointed by this future.

Aside: I can't stand popups when I'm coding. I always have to change it to something more manual or it just breaks up my flow. ^K to bring up a man page from vim is great--that's what I'm after.

There are dozens of us!

I never understood how people could maintain focus while their cursor is jumping around and things are flying across the screen. I can type 90+ wpm, I don't need help typing.

I think there's a fundamental difference.

Autocomplete and intellisense are tools first and foremost. AI is centralized into a handful of companies you are forced to pay every month for.

Autocomplete and intellisense don't care about your data. There's an inherent issue with data, privacy and ownership when it comes to AI.

If we can run useful models locally and make it generally available on consumer hardware... things would be different.

Reply on news.ycombinator.com