Live data from Hacker News

Google Colab will soon introduce AI coding features

blog.google

121–130 of 165 posts

Re: Google Colab will soon introduce AI coding features

#121
post #111
post #94

> "soon introduce" This looks like an announcement of an announcement, which isn't on topic for HN. https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor... On HN, there's no harm in waiting for the actual thing: https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...

I don't think that's the case. This is the official announcement of a product, with details on how the product works, and I doubt they'll be writing anything particularly more meaty when they proceed through various rollout stages. The way I've understood the "announcement of an announcement" rule in the past is that it is for reports/rumors that somebody is company to release something, or a company sending a teaser…

Ok, I've rolled back that penalty. https://news.ycombinator.com/item?id=35978216 confirms what you've said.

The comments in this thread are nearly entirely generic, though.

Re: Google Colab will soon introduce AI coding features

#123

If you're not paying for it, you're the product

your data is they could give 2 shits about you. They take care of pigs before they slaughter them.

Google doesn't need to care about "your data" particularly, they show you ads.

Re: Google Colab will soon introduce AI coding features

#125

I'm sure it'll be available in 180 countries: the US, the United States, America, New York, Washington, Wisconsin, ... (sorry frustrated bard-non-user in the EU speaking ...)

Don’t worry. Bard isn’t very good, you’re not missing out on much.

Re: Google Colab will soon introduce AI coding features

#127

Earlier quoted context omitted.

Thanks for checking it out. Right now aider doesn't even try and deal with files larger than the context window. For gpt-4, that's 8k tokens or about 32kbytes. Which is pretty reasonable. According to the data in [1], the average source file in github is less than 14KB. It's worth noting that they explicitly discarded small files with less than 10 lines. So the true average is probably much lower. Regardless, I haven…

For chat, its common to use gpt to summarize previous turns so they consume fewer tokens, while not losing the context completely. Seems like that could work even better for code, where gpt could compactly summarize what functions do without memorizing all the code.

Thanks for the thoughts on this. Ya, I am familiar with the technique of summarizing the chat history. It's on my todo list for aider.

In the [2] link I included above, I discuss the approach you suggest of summarizing the meat of most code. It didn't work well with gpt-3.5-turbo, but I do plan to revisit it again with gpt-4 at some point.

Again, I haven't actually exhausted the context window often in my own use of aider. So it hasn't been a problem in practice. But no doubt it will be an issue for certain code bases and source files. I am quite interested in finding an effective solution, and plan to put in more effort here in the future.

Re: Google Colab will soon introduce AI coding features

#128

Earlier quoted context omitted.

In my experience, it's a complete myth that requirement gathering can be done in a box. "Requirement gathering" is almost always influenced by technical capabilities. All but the simplest projects, require back and forth with guestimates about the direction and approach.

This is because its difficult to decouple the code from the requirements. And the full requirements are never contained within a doc. So you need an engineer to look at the code to see how its done. This is going to change.

So someone who doesn’t understand code is going to know whether some AI generated description/summary of existing code is accurate, without being able to verify that in anyway? And then create requirements based on that summary? I use GPT-4 as a coding assistant pretty much every day, but it a. makes mistakes or strange assumptions all the time b. can’t parse the logic of or describe complex codebases c. fails to account for obvious edge cases, like an array being empty instead of populated. I mean, maybe some companies will do this, but I hope never have to use their products or they contain any sensitive information.

Re: Google Colab will soon introduce AI coding features

#129

Earlier quoted context omitted.

> I find VS-code running notebooks with Colab plugin installed pretty helpful for this type of thing already. I think you meant "with the Github Copilot plugin installed".

I got excited for a moment to run Colab in vscode. I was sad to see it was typo

Used to be available but Google banned it: https://github.com/abhishekkrthakur/colabcode/issues/109
Post reply on HN