you are viewing a single comment's thread
view the rest of the comments
[–] 30 points 13 hours ago* (7 children)

In theory, LLMs can help people do better quality work faster, but the productivity boost is modest when quality work is the goal. However, what seems to be happening to a great extent is that people are putting in a lot less effort and employers are now paying for salaries and AI tokens while getting worse quality work and slower progress in return.

And yeah, anyone who actually is trying to do a good job will get pissed at their co-workers half-assing it and pumping out low-effort AI slop while pawning work off on them. I think the only sustainable path forward with AI is instead of token usage being a badge of honor, there’s a culture of shame around AI where people put in a lot of effort to make sure nobody can tell they used AI out of fear of ridicule.

  • source
  • hideshow 7 child comments
  • [–] 12 points 11 hours ago

    there’s a culture of shame around AI where people put in a lot of effort to make sure nobody can tell they used AI out of fear of ridicule.

    Earlier this year there was a whiteboard in one section of my office where some people had a tokenmaxxing leaderboard. Now it suddenly disappeared.

  • source
  • parent
  • [–] 11 points 10 hours ago (2 children)

    I use ai at work to handle low-value, high-effort bullshit.

    I have to plan out a quarter ahead for my team of about 20 people. The expectation was to sit with them during multiple planning meetings, haggle out plans based on estimates and vibes for what we might be able to get done.

    I hated that so I loaded all of our work data (agile) into an agent I created. I had it create plans based on historical data. I tuned those plans based on what I needed to. It would take me about two hours to do the entire quarterly planning. I sent it to the teams for modifications, fed those back through for final tuning. Then the agent would create and schedule all of the work, dates, assignments, set up all the admin stuff to make it go.

    Took planning from probably about 60 labor hours per quarter to maybe 10.

    Because my boss was a twat, he would not log into the system to view the plans, he could only read PowerPoint. I had the ai make his dumb ass slides from the content.

    So you can use these things effectively. You can also make endless work slop. Keep in mind, work slop is not new. Ever hear of a "walking deck"? It is as fucking stupid as it sounds. You make a presentation about what you and your team does for nobody in particular. It is usually self promotional marketing wank that nobody ever looks at. Been a thing for ages. Now I can just use my work data to generate one if I wanted (I don't want one).

    This levels the playing field for all the busybodies who don't really work, but do make a lot of slop. I hope they drown in it.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 3 points 10 hours ago

    Yeah, LLMs are useful for delegating work that is labor intensive and bullshit in nature. I bet Peter from Office Space would have no problems with his TPS reports if a LLM were helping out.

  • source
  • parent
  • [–] 1 point 23 minutes ago

    My frustration is that no one is validating AI output, at least on my team. We've mostly been working on CVEs recently. I won't deny that AI has accelerated the process, in that instead of spending large amounts of time researching problems and applying the fixes manually, I can just copy and paste the answer it provides ... But generally the answer is text like "updated the vulnerable package," even though we haven't (because usually the updated package isn't available, to continue this example).

    Many on my team are experienced and competent and even those people are just relying on the CI pipeline to validate for them, but so far as I can tell it just checks to see whether the vulnerability is mentioned in the configuration files.

    Some of the team members are new. I'm 100% certain they're just copying and pasting.

    I am experienced. I was trying to do the job right by validating before I made entries, but I was repeatedly chastised for being slower than the people who are just copying AI output. My company hasn't outright said but has heavily implied "if the customer [who provides the validation pipeline] doesn't care, why should we?"

  • source
  • parent
  • [–] 5 points 11 hours ago (2 children)

    Or they could just actually not use it?

    Using it and then obfuscating it’s use is just creating more unnecessary work to end up with a lower quality finished product.

    It’s not really that difficult to just not use it, I’ve gone literally my entire life without it and nothing has been “too difficult”.

    Thousands of generations have passed before me without using it.

    There is no element of inevitability to this. You can just not use it.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 4 points 9 hours ago (1 child)

    But then I'd have to think about the task, instead of just rephrasing a few things.

    Most people didn't really care about their jobs before LLMs, and now management has basically given them permission to show it overtly. If it's a bullshit waste of time, then just have the LLM do it. Nobody cares, nothing about our economy has been real since 2008.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 3 points 5 hours ago

    There are people who should care

    If you have a bullshit job then fuck it (except you’re still damagibg both the environment and the economy just to be lazy) but I’m talking about people in roles that are actually useful to society.

  • source
  • parent