Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)T

TehPers

@ TehPers @beehaw.org

Posts
0
Comments
624
Joined
3 yr. ago

  • Block ditches 4,000 staff, because AI can do their jobs

    Jump

  • In some cases, it appears to be the opposite: CEOs want to do mass layoffs, so they blame AI rather than taking accountability themselves. The Amazon layoffs reek of this.

  • A 25-Person Startup Built a Chip That Only Runs One AI Model. It's 73 Times Faster Than Nvidia.

    Jump

  • had people understood from the start the limitations of it, investment would've been more modest and cautious

    People did understand from the start. Those who do the investing just didn't listen, or they had a different motive. These days it's impossible to tell which.

    And by "people" I'm not referring to random people, but those who have been closer than most to the development of these models. There has been an unbelievable amount of research done on everything from the effectiveness of specific models in niche fields to the ability to use an LLM as the backend for a production service. Again, no amount of negative feedback going up the chain has made a difference in the direction, so that only leaves a few explanations on why the investment continues to be so high.

  • Is it a bad practice to replace compiler warnings with a bunch of TODO notes?

    Jump

  • Could also do this:

     rs
        
#[expect(lint, reason = "TODO: #issue")]

    Edit: to clarify, #issue is an issue number that points to a related issue or task. Could also just explain it inline, but if you have a task tracker, better to make a task instead.

  • Block ditches 4,000 staff, because AI can do their jobs

    Jump

  • Not exactly. Thinking models just inflate the context window to point the model closer to your target. GANs have two models which compete against each other, both training each other, with the goal of one (or both) of those models being improved over time.

  • Block ditches 4,000 staff, because AI can do their jobs

    Jump

  • This is unironically what I've seen people try to do, except they assume the second AI is correct.

    Unrelated, but this is how GANs work to some extent. GANs train during the back-and-forth though, while LLMs do not.

  • Block ditches 4,000 staff, because AI can do their jobs

    Jump

  • Which outputs are accurate, and which ones are inaccurate? How could you tell? What steps did you take to verify accuracy? Was verifying it a manual process?

  • Uiua — an extremely terse programming langauge

    Jump

  • Is this an esolang? Could be nice for code golf maybe.

  • A 25-Person Startup Built a Chip That Only Runs One AI Model. It's 73 Times Faster Than Nvidia.

    Jump

  • Shh you'll pop the bubble if you start talking sensibly. It's not an ASIC—it's a specialized piece of hardware optimized to execute a model with unparalleled performance. Now buy my entire stock of them and all the supply for the next two years please.

    (Figuring out the compose combination for an emdash took longer than I'd like to admit lol)

  • How to do a collaborative API cache?

    Jump

  • Can't speak for Git, but caching responses is a common enough problem that it's built into the standard HTTP headers.

    As for building a cache, you'd want to know a few things:

    • What is a cache entry? In your case, seems to be an API response.
    • How long do cache entries live? Do they live for a fixed time (TTL cache)? Do you have a max number of cached entries before you evict entries to make space? How do you determine which entries to evict if so?
    • What will store the cache entries? It seems like you chose Git, but I don't see any reason you couldn't start simple just by using the filesystem (and depending on the complexity, optionally a SQL DB).

    You seem locked into using Git, and if that's the case, you still need to consider the second point there. Do you plan to evict cache entries? Git repos can grow unbounded in size, and it doesn't give you many options for determining what entries to keep.

  • Firefox’s AI Kill Switch is a Trap: How Mozilla Made AI Your Problem

    Jump

  • Pre-LLM translation services also generally used AI, just via more traditional machine learning. The only difference is introducing a locally run LLM.

    If it runs locally and is openly available, then it doesn't make much difference to me if it's a traditional model or a LLM.

  • What will happen if the Linux kernel starts having AI generated code in it?

    Jump

  • Do we know it doesn't?

  • AIs can’t stop recommending nuclear strikes in war game simulations | OpenAI, Anthropic and Google opted to use nuclear weapons in simulated war games in 95 per cent of cases

    Jump

  • Terminator might be a little more popular.

    It seems the only way to win is not to play.

  • The Junior Developer Is Going Extinct

    Jump

  • For what it's worth, open source dev can also work. If you can commit some time to a project you care deeply about and make regular contributions, that's another form of experience, and I see no reason you couldn't add that as a line to your resume alongside any other work experience.

  • The Junior Developer Is Going Extinct

    Jump

  • This has always been an issue. From my experience, the best way to get in was through internships, co-ops, and other kinds of programs. Those tend to have lower requirements and count as experience.

    Of course, today, things are a lot different. It's a lot more competitive, and people don't care anymore about actual software dev skills, just who can churn out SLOC the fastest.

  • California’s New Bill Requires DOJ-Approved 3D Printers That Report on Themselves

    Jump

  • *Several states

    Washington, for example, has a similar bill proposed.

  • Amazon blames human employees for an AI coding agent’s mistake | Two minor AWS outages have reportedly occurred as a result of actions by Amazon’s AI tools

    Jump

  • To put some perspective into what our code looks like, there are very few tests (which may or may not pass), no formatter or linter for most of the code, no pipelines to block PRs, no gates whatsoever on PRs, and the code is somewhat typed sometimes (the Python, anyway). Our infrastructure was created ad-hoc, it's not reproducible, there's only one environment shared between dev and prod, etc.

    I've been in multiple meetings with coworkers and my manager talking about how it is embarassing that this is what we're shipping. For context, I haven't been on this project for very long, but multiple projects we're working on are like this.

    Two years ago, this would have been unacceptable. Our team has worked on and shipped products used by millions of people. Today the management is just chasing the hype, and we can barely get one customer to stay with us.

    The issue lies with the priorities from the top down. They want new stuff. They don't care if it works, how maintainable it is, or even what the cost is. All they care about is "AI this" and "look at our velocity" and so on. Nobody cares if they're shipping something that works, or even shipping the right thing.

  • Amazon blames human employees for an AI coding agent’s mistake | Two minor AWS outages have reportedly occurred as a result of actions by Amazon’s AI tools

    Jump

  • Colleagues, and the issue is top-down. I've raised it as an issue already. My manager can't do anything about it.

  • Amazon blames human employees for an AI coding agent’s mistake | Two minor AWS outages have reportedly occurred as a result of actions by Amazon’s AI tools

    Jump

  • Because if I spent my whole day reviewing AI-generated PRs and walking through the codebase with them only for the next PR to be AI-generated unreviewed shit again, I'd never get my job done.

    I'd love to help people learn, but nobody will use anything they learn because they're just going to ask an LLM to do their task for them anyway.

    This is a people problem, and primarily at a high level. The incentive is to churn out slop rather than do things right, so that's what people do.

  • Amazon blames human employees for an AI coding agent’s mistake | Two minor AWS outages have reportedly occurred as a result of actions by Amazon’s AI tools

    Jump

  • This is what happens to us. People put out a high volume of AI-generated PRs, nobody has time to review them, and the code becomes an amalgamation of mixed paradigms, dependency spaghetti, and partially tested (and horribly tested) code.

    Also, the people putting out the AI-generated PRs are the same people rubber stamping the other PRs, which means PRs merge quickly, but nobody actually does a review.

    The code is a mess.