top 97 comments

sorted by: hot top controversial new old
[–] 207 points 4 months ago (12 children)

So typing isn't fast enough to burn all the daily tokens?

  • source
  • hideshow 12 child comments
  • [–] 93 points 4 months ago (4 children)

    The faster you burn them the sooner you knock off for the day.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 73 points 4 months ago (4 children)

    Techbros will have workers do anything but work from home.

  • source
  • hideshow 4 child comments
  • [–] 29 points 4 months ago (3 children)

    Or have single-person offices instead of an open space

  • source
  • parent
  • hideshow 3 child comments
  • [–] 103 points 4 months ago* (6 children)

    in this one picture, I see generations of so many individual great ideas coalescing into one fantastically bad idea (and stupidly comical consequence) that is so... bizarre that I can both simultaneously understand why nobody really saw it coming, and am in low-key disbelief that this is even real...

  • source
  • hideshow 6 child comments
  • [–] 30 points 4 months ago* (2 children)

    I mean, that's a steno mask, and anyone who's had issues with hand pain but wants to communicate via text has probably wished for something resembling this. The problem is that they're obnoxiously expensive. (And they look ridiculous, but that's its own issue.)

    When I'm in public, I wish people had these. I neither need nor want to listen to your phone conversations.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 2 points 3 months ago (1 child)

    Their price is so dumb, that thing shouldn't cost that much

    Just like those accessibility tools that let you use control the mouse pointer using only your mouth etc etc, they have stupid prices when some aren't even that complex

  • source
  • parent
  • hideshow 1 child comment
  • [–] 1 point 3 months ago

    They're a niche product that has a very solid use case (court transcripts). They require far less training than a stenotype, but because they're so niche, the people who do make them can more-or-less charge what they want. If we're being less skeptical, the low demand means the manufacturers can't take advantage of the efficiencies of scale, so the price remains high.

  • source
  • parent
  • [–] 7 points 4 months ago* (last edited 4 months ago) (2 children)

    of course it’s not, because every laptop and airpod has noise cancelling input isolation

    But as far as rage bait goes, top tier

  • source
  • parent
  • hideshow 2 child comments
  • [–] 35 points 4 months ago (13 children)

    Did the CEOs unplug their keyboards? Wtf even is this?

  • source
  • hideshow 13 child comments
  • [–] 32 points 4 months ago* (12 children)

    Well the mask is a steno mask

    Theoretically most people will speak faster than they type. You have to type around 180-200 wpm to be faster than speaking.

    (I say theoretically, because usually typing speed ratings also ding you for errors, and uh, speech transcription isn’t really there, either.)

  • source
  • parent
  • hideshow 12 child comments
  • [–] 27 points 4 months ago (3 children)

    The hard part of both speech and typing is thinking about what you say. Typing nor speaking are going to change the speed I can get information into the computer.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 10 points 4 months ago (2 children)

    Maybe we could ask the AI to do that thinking bit then tell us what to say.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 1 point 4 months ago (7 children)

    I'm not really into it but one of the guys on late night linux podcast, generally resistant to LLMs and shiny new things, is a fan of speech to text for general computing, as in every input field in the OS should support speech. In a recent episode he said that he believes it to be the way of the future.

  • source
  • parent
  • hideshow 7 child comments
  • [–] 3 points 4 months ago* (6 children)

    I remember Dragon SpeechDragon Naturally Speaking saying the same thing in the 90's. It's improved, but not enough to make it useful as more than an aide for people who can't type. I do agree, that for simple accessibility, it should be integrated into every field, but I doubt it's ever going to take over.

    As others have noted, that it's only technically true that dictation is faster than typing. In a practical sense, there's a fair number of reasons why that's not the case, including that usually thinking about the entry is what's the slowest, and also the errors in both are typically what slows people down.

    there's also the problem of, for example, keeping entries confidential. You don't want to speak your passwords where others can hear you.

  • source
  • parent
  • hideshow 6 child comments
  • [–] 3 points 4 months ago

    I remember Dragon! And ViaVoice! I saw a presentation for ViaVoice in the late 90s and it blew my tiny mind.

    It really was the future. And it’s… a bit better since then. Oh god thats like 30 years ago almost

  • source
  • parent
  • [–] 1 point 4 months ago (4 children)

    It’s improved, but not enough to make it useful as more than an aide for people who can’t type.

    I don't think this is true.

    There's a locally hostable model called whisper that is very impressive.

    My plumber uses speech to text to send text messages all day.

    Late Night Linux guy says he uses it for microsoft teams quite a bit.

    You're only partially correct about input speed. If you want to dictate an email then yes you need to think about each word you want to say and the order in which to say them. Coupled with an LLM that problem is diminished because you can just kind of have a conversation with the LLM and tell it to draft an email.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 2 points 4 months ago (3 children)

    You’re only partially correct about input speed. If you want to dictate an email then yes you need to think about each word you want to say and the order in which to say them. Coupled with an LLM that problem is diminished because you can just kind of have a conversation with the LLM and tell it to draft an email.

    and how much of that conversation with an LLM is "No, what I want is..." because it assumed something; or just straight up hallucinated or the typo made it go off on a tangent?

    As for whisper, I can find sources that are saying for American-English speakers in a not-noisy environment (aka the best case scenario,) the model has a word error rate between 2-8%. For reference, Dragon NaturallySpeaking had a WER of 3-5%. So I wouldn't say that Whisper has made any substantial improvements, and they're OpenAi. you can trust them if you want. I don't think that'll work out well in the long run, though.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 1 point 4 months ago (2 children)

    I'd like to see the source that says Dragon's WER in the 90s was 3-5%. I used Dragon in the 2000s and it just wasn't comparable to the current state of the art.

    whisper.cpp is an opensource implementation, although I'm not certain exactly how open.

    when you're providing context rather than instructions the tendency for a model to hallucinate or run off on a tangent is minimal, because the context you're providing has it's own cohesion.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 1 point 4 months ago (1 child)

    I’d like to see the source that says Dragon’s WER in the 90s was 3-5%. I used Dragon in the 2000s and it just wasn’t comparable to the current state of the art.

    https://dragon-medical-transcription.com/history_speech_recognition.html, for example. a lot of adverts and awards were given to it (admittedly awards like PC Mag that were probably paid advertising... but that's why I went with Open AI's assessment on whisper at 2%.) Dragon was boasting 99% accuracy after (admittedly months) of training; and it frequently reached it. there were some gotchas in that- the months-long training was a big one. The other was that you frequently had to slow down and be careful to enunciate that you don't have to do with modern systems (including the MS versions of Dragon- they bought it out at some point)

    whisper.cpp is an opensource implementation, although I’m not certain exactly how open.

    It's on the MIT license, if that helps. I take issue with anything OpenAI is involved in. for oh-so-many reasons.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 34 points 4 months ago (3 children)

    Hook it up to a bong and you've suddenly made work a hell of a lot more interesting.

  • source
  • hideshow 3 child comments
  • [–] 33 points 4 months ago (4 children)

    Anyone needs handmade wooden furniture?

    Because that's what i'm going to be switching careers to if that trend comes to my place.

  • source
  • hideshow 4 child comments
  • [–] 22 points 4 months ago (2 children)

    Woodworking is surprisingly popular among tech folk. It seems some hobbies just click better for techies. Bouldering is another example.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 10 points 4 months ago (1 child)

    Sewing/designing clothes really clicked for me. Haven't tried woodworking, but I imagine it scratches the same itch and utilizes similar skills: 90% of sewing is just planning, calculation, and measuring. Then, watch everything just fit together into place

  • source
  • parent
  • hideshow 1 child comment
  • [+] 27 points 4 months ago* (5 children)
  • [–] 3 points 4 months ago (4 children)

    how does a headset isolate their voice

  • source
  • parent
  • hideshow 4 child comments
  • [+] 6 points 4 months ago* (last edited 4 months ago) (2 children)
  • I imagine it's like the Active Noise Canceling (ANC) mode of headphones which will monitor background noise and then calculate the inverse sound wave and play that sound to you which in effect mostly cancels out the background noise (works better for certain frequencies than others). For a microphone they could do the same but instead of playing you the inverse sound, they could play it through the microphone for whomever is listening.

  • source
  • parent
  • [–] 25 points 4 months ago (1 child)
  • [–] 21 points 4 months ago

    This is such a Kafkaesque cyberpunk dystopian problem. This is needed because owners and managers want open offices for surveillance and coat cutting. They are also aware that only about half of workers thrive in those environments while the other have drops in productivity. Solution a cheap muzzle so that the workers who aren't as productive in noisy environments aren't disturbed.

    Meanwhile could you imagine the smell of those things. You'd have to wash them everyday or have some kind of cover for them not to just have a permanent funk.

  • source
  • [–] 40 points 4 months ago (4 children)

    Why stop at feeding? Where's the penis and vagina tubes? We might as well be jorkin' it too if all the rich and powerful are.

  • source
  • hideshow 4 child comments
  • [–] 38 points 4 months ago (9 children)

    Or, listen me out, they could work from home.

  • source
  • hideshow 9 child comments
  • [–] 39 points 4 months ago (8 children)

    Working from home doesn't appeal to the emotional needs of fragile managers.

  • source
  • parent
  • hideshow 8 child comments
  • [–] 9 points 4 months ago (7 children)

    It doesn't appeal to the emotional needs of a bunch of my colleagues who are in the office every day voluntarily either.

  • source
  • parent
  • hideshow 7 child comments
  • [–] 13 points 4 months ago*

    Yeah, I have colleagues who choose to work in the office when work from home is available because they like the separation of work from home, don't have a good spot to work from home, are aware they would be distracted at home, prefer to see other people in person, and a bunch of other reasons. At least they get a choice!

  • source
  • parent
  • [–] 10 points 4 months ago (5 children)

    Every day I'm a little surprised there's no news story of some workers beating their "no, you have to come into the office" manager to death. They've got means, motive, and opportunity, and it's extra funny because if they'd been allowed to work at home they wouldn't have at least two of those.

    But really we're ruled by the worst of us. Cowards and fools.

    Maybe unionizing is safer than hitting the decision makers with an office chair while screaming "you made this possible" until they can't even cry anymore.

  • source
  • parent
  • hideshow 5 child comments
  • [–] 5 points 4 months ago (4 children)

    Well what I'm trying to tell you is that there are probably more people than you realise who want to be in the office. My partner, and a bunch of my coworkers, hated being forced to work from home during the pandemic. So maybe that's part of the reason.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 4 points 4 months ago (3 children)

    Oh, I read your thing backwards then.

    I can't imagine wanting to go into the office on the regular. The commute. The lost time (can math out to like a 20% pay cut, if you spend two hours a day traveling + getting ready). The sickness. The lack of control over environment (temperature, sound).

    Can't relate to it. And I'm a very social person that likes interacting with people.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 1 point 4 months ago (2 children)

    I live 15 minutes from work. If I got out of bed early enough I could easily bike there (and have done so).

    Really the only thing I miss from work from home is the ability to take a short break once per hour to do some bodyweight exercises or kettbell swings during lunch.

    But on the flipside, it's hard for me to stay concentrated while at home. I personally get way more work done in the office.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 1 point 4 months ago (1 child)

    But on the flipside, it’s hard for me to stay concentrated while at home.

    While I believe that is true for you, I don't believe it justifies the lost time, health, and environmental damage of mandating in-office for everyone.

    The office is super distracting for many people.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 54 points 4 months ago (18 children)

    Isn’t it cheaper and quieter to just type out your prompts?

    This is akin to people who have conversations on speakerphone in public places.

  • source
  • hideshow 18 child comments
  • [–] 4 points 3 months ago (4 children)

    Older people such as myself tend to hate voice-to-text I think because it was so awful in the past. And if you screwed up with it in the past it was a less understandable excuse that “I was using voice-to-text.” And because we were all forced in some way to learn to type well.

    Voice to text works a little bit better now. And I think younger people know everyone else uses it and to forgive when it screws up.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 1 point 3 months ago

    It worked better five years ago before they replaced the existing algorithms with AI bullshit. Keeps adding slurs to my dictionary too since they replaced keyboard prediction with modern AI and so many people use slurs Google just assumes I do too.

  • source
  • parent
  • [–] 4 points 4 months ago (4 children)

    apparently audio and images are more efficient compared to text for multimodal models?

  • source
  • parent
  • hideshow 4 child comments
  • [–] 27 points 4 months ago (2 children)

    I'm going to need significant levels of convincing. Computers have always preferred specificity and accuracy, it's half the reason I'm in my current position (MSP Escalations/level 3, half of my success at fixing issues is being extremely specific in looking up exact error messages instead of paraphrasing).

    This isn't a defense of AI; on the contrary, it's my doubt that AI can read intentions/inflection/emotion better than just writing out what you actually want.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 8 points 4 months ago* (last edited 4 months ago)

    Deepseek recently published a paper in which they describe that vision tokens contain more information than text tokens and that this can be used to compress context.

    We present DeepSeek-OCR as an initial investigation into the feasibility of compressing long contexts via optical 2D mapping.

    Experiments show that when the number of text tokens is within 10 times that of vision tokens (i.e., a compression ratio < 10×), the model can achieve decoding (OCR) precision of 97%. Even at a compression ratio of 20×, the OCR accuracy still remains at about 60%. This shows considerable promise for research areas such as historical long-context compression and memory forgetting mechanisms in LLMs.

    It reminds me of LLM caveman speak, it used to have another option to use Chinese instead of English. A language like Chinese is seemingly better at encoding information in fewer tokens and I think this is the same mechanism why OCR tokens work so well.

    That said, I also doubt that voice messages are more efficient than text prompts, but it's best not to waste too much time engaging with these sorts of LinkedIn posts (and LinkedIn in general).

  • source
  • parent
  • [–] 18 points 4 months ago* (last edited 4 months ago) (1 child)

    How about we just get to the logical conclusion:

    ✨ Techbro Land ✨

    Hook them up to a simulation, yes, like The Matrix! In this simulation they get to do everything they dream of.

    Crypto shitcoins as far as the eye can see. Their stonks all go to the moon. Everybody is prompting Ai any time their neurons accidentally bump together. The simulation will accurately simulate an entire earth population just prompting AI and glued to their phones. Look! More profits! Wow.

    Ai gambling! Ai sports! Ai art! Ai dating! Ai social media! Ai business! Ai porn! It's all here in ✨ Techbro Land ✨.

    We can even accurately simulate mass misery of the general population so our little tenants feel superior! Meanwhile, the Techbro doesn't need to sleep. They believe they unlocked the code to Immortality by micro dosing various hallucinogens and stimulants, allowing them to spend 100% of their day in a chair reviewing an Ai agent's reviewing of vibe code!

    There are no "regulations" or even governments in ✨ Techbro Land ✨ . There's a single AI Simulated Singularity (A.S.S) TechBroKing Fascist that makes all the memes, and punishes all the poor working consumption class. Every living situation is AirBnB. All the cars are self driving 24 ton pickups. Trains only exist to ship drones and supplies for infinite data centers and government contracts.

    Computing is subscription only, games are rented by the minute, nobody owns anything.

    Profits go up and up and up and up and up without end. No climate change, despite the global temperature going up 12⁰C, wars are fought daily with killer robots...I could go on.

    The important thing is: These little hoodie-clad megalomanics can't even tell they're actually just locked safely away from the rest of us in stasis in a former mega data center.


    ...The rest of us, in the real world, take the best of our computing standards with us, and dial all the bullshit back to like 2004 and try this shit again.

  • source
  • hideshow 1 child comment
  • [–] 18 points 3 months ago (2 children)

    This looks uncomfortable and humiliating. Now if they were to make it in the form of a suppository...

  • source
  • hideshow 2 child comments
  • [–] 36 points 4 months ago
    [–] 49 points 4 months ago

    I would rather do almost anything than talk to a device, except in very specific circumstances.

    I set timers and play music on a smart speaker somewhat often.

    Occasionally, when I am alone, know exactly what I want to say, and my hands are full, I might dictate a text message.

    But other than that, I will not be talking to my device, thanks. The human voice is primarily for talking to other humans, with all the imprecision and uncertainty and emotional resonance that entails. Keyboards are great tools designed for precise computer input, and I would like to continue to use them.

  • source
  • load more comments
    view more: next ›
    26 comments in c/programmerhumor@lemmy.ml view that post