GraphQL man… the whole thing does this by design. Just in the last week I had to implement a custom retry function because the python requests one very reasonably doesn’t consider a 200 status code to be an error.

Yeah, this is the problem with frankensteining two systems together. Giving an LLM a prompt, and giving it a module that can interpret images for it, leads to this.

The image parser goes “a crossword, with the following hints”, when what the AI needs to do the job is an actual understanding of the grid. If one singular system understood both images and text, it could hypothetically understand the task well enough to fetch the information it needed from the image. But LLMs aren’t really an approach to any true “intelligence”, so they’ll forever be unable to do that as one piece.

Eh, this is a thing, large companies often have internal rules and maximums about how much they can pay any given job title. For example, on our team, everyone we hire is given the role “senior full stack developer”, not because they’re particularly senior, in some cases we’re literally hiring out of college, but because it allows us to pay them better with internal company politics.

I don’t necessarily disagree that we may figure out AGI, and even that LLM research may help us get there, but frankly, I don’t think an LLM will actually be any part of an AGI system.

Because fundamentally it doesn’t understand the words it’s writing. The more I play with and learn about it, the more it feels like a glorified autocomplete/autocorrect. I suspect issues like hallucination and “Waluigis” or “jailbreaks” are fundamental issues for a language model trying to complete a story, compared to an actual intelligence with a purpose.

Same honestly. And if I ever ask a question that someone might think is a duplicate, I link to that question and say something like “I found X, but the answers here don’t reflect Y”.

I’d be very surprised if anything functional actually comes out of this. Far more likely they get scammed out of the money by garbage like the current “AI writing detection” methods, with terrible success rates that cause more societal problems than they solve.

Surprisingly legible, but feels like I can only read it with momentum, flitting past it and letting my subconscious tell me where the word breaks are. The moment I get confused and look more closely, it becomes almost impossible to read.

Exactly the mistake threads just made, trying to capitalize on twitter’s rate limiting fiasco. The “general public” is extremely fickle, and Reddit will give us more opportunities.

I also feel like a lot of the value of chronological is lost if I think it’s algorithmic recommendations. If I don’t know I’m browsing the latest? I’ll likely just think the algorithm is serving up some garbage. Especially somewhere like Facebook, where people haven’t really been curating their feed for years, just… following whoever to be polite and letting the algorithm take care of it.

TL;DR: People like searching for answers to their question from Reddit, since it’s got upvotes and a lot of subreddit answers to niche opinion questions. Reddit is one of the most popular search keywords, and is unreliable right now.

Also recaps several other recent Reddit headlines, which I’m sure you’ve seen if you’re hanging out in this subreddit.

Don’t love the framing of this paragraph from TechCrunch. It’s not that they’re charging for the API. That’s understandable and obvious, and we all wanted the platform to survive. I’ll be happy to volunteer to contribute to lemmy development/server costs/app development one day. It’s that they’re grossly overcharging for the API to such an extreme degree that paid subscriptions to third party apps actually lose money.

In April, Reddit announced its plans to start charging developers to access data through its API. The move was obvious — to restrict third parties from accessing Reddit data that can help build text-generating machine learning models such as OpenAI’s GPT 4. Developers building apps and bots to assist people using Reddit and researchers who wish to study the platform for noncommercial purchases were among the few exceptions. However, as a result, third-party apps, including popular Reddit client Apollo, found it difficult to pay for those charges and decided to go offline. Various popular subreddit moderators came in support of those apps and developers and started protesting against the API pricing move.

Oof, so counterproductive. I’m a hard reviewer, always try to hold others to the standard of code I’d like to work in, and be held to myself, but every once in a while I see a PR that’s just… no changes required.

I love just hitting accept without making any feedback, it means my coworker valued my feedback and actually internalized it. Trying to laser in and nitpick something unnecessary would be a waste of all our time.

I used an extension a while ago that changed CSS colour values (#ababab) into little coloured dots, that became a colour picker when clicking on them (while still letting you input RGB or Hex, ofc), and it was pretty awesome!

So, I could unironically see this being really nice. Although… I think this would need a pretty narrow context, something like if x == true would look pretty confusing as a toggle, I imagine. But assigning x = true? Bring it on.