• 0 Posts
  • 254 Comments
Joined 2 years ago
cake
Cake day: June 16th, 2023

help-circle


  • judgement

    Yeah, it admitted to an error in judgement because the prompter clearly declared it so.

    Generally LLMs will make whatever statement about what has happened that you want it to say. If you told it it went fantastic, it would agree. If you told it that it went terribly, it will parrot that sentiment back.

    Which what seems to make it so dangerous for some people’s mental health, a text generator that wants to agree with whatever you are saying, but doing so without verbatim copying so it gives an illusion of another thought process agreeing with them. Meanwhile, concurrent with your chat is another person starting from the exact same model getting a dialog that violently disagrees with the first person. It’s an echo chamber.




  • Comment resonates with my experience.

    Software project at work recently:

    We are going to launch a new offering to improve experience for customers.

    Ok, how?

    We are going to switch it to cloud model and charge annually instead of perpetual.

    Ok, that’s for us, what about customer?

    We are going to analyze their accounts and present them with suggestions on other of our products and addons they haven’t bought yet.

    Where is the customer improvement?

    We are going to discontinue supporting third party products and focus exclusively on customers that buy only from us.

    Ok, but we have support for third party products we don’t even compete with?

    We are going to exclude those too, to focus on the market that is important.

    Ok, but at least you’re going to provide equivalent capability as the product you are replacing?

    We are going to streamline the experience by offering only the core capabilities and discontinue extraneous features.

    Ok, but you think this will expand revenue, so you will afford to explain the service and support team and free up more time for developers to get requirements?

    We are in fact going to lay off and offshore all of it, including most of the customer contacts that barely kept the preceding product alive…

    Now after a while of this mess they also had like 96% availability with almost all of it unplanned outages, but that’s not too bad because they have only like 6 or 7 customers anyways. There’s emails running around asking why the product has failed, and the answer seems to be we need to kill more of our successful products to try to push customers into this mess.



  • Yeah, already things were getting harder to follow as people went to address the “strangely sparse cities” problem by flooding the environment with way more stuff aiming for more plausible, but it’s more than you can ever consume and it’s generally hard to know when you are actually supposed to pay attention or not. Finding interesting side quests among the flavor text used to be a thing, but now the flavor text is just overwhelmingly too much for that.

    Of course, there’s recognition of that and games start putting indications of “THIS RANDOM NPC HAS SOMETHING TO SAY” bright over anyone vaguely important. So I suppose in that context NPC flavor text vomit might as well be AI since it’s been clearly indicated as stuff to ignore as background noise. Still disappointed in the decline of “is this important or not” determination being organic.





  • On the other hand, my news feed keeps offering articles from “people” telling me how much better it is to rent and it’s a sucker"s game to own. Usually it’s based on an overstated claim of convenience, but most recently one tried to use all sorts of hand waving to assert that it is also financially better. Extra rich that the author even admits he owns a lot of property that he rents out while simultaneously trying to make it sound stupid to actually own your own house (claimed he himself rented so that he’s not “stuck” paying his own loan). I’m sure these are very sincere and smart people, why would they steer anymore wrong?

    Don’t know if it’s real or my imagination, but it seems there are times when there’s extra pressure to convince people that they want to eternally rent.





  • Most people I know haven’t even bothered to buy a new TV since Dolby Vision was created. A fair number still have 1080 sets.

    While some like you may certainly demand it and it would be a good idea, I think it’s a fair description to help people understand the goal is an android TV like experience, and a lot of people are oblivious to a lot of the details of picture quality.

    Just a bit over the top for such an overly dismissive statement, versus saying something like “does it support Dolby vision? I won’t be interested until it does”





  • The issue here is that we’ve well gone into sharply exponential expenditure of resources for reduced gains and a lot of good theory predicting that the breakthroughs we have seen are about tapped out, and no good way to anticipate when a further breakthrough might happen, could be real soon or another few decades off.

    I anticipate a pull back of resources invested and a settling for some middle ground where it is absolutely useful/good enough to have the current state of the art, mostly wrong but very quick when it’s right with relatively acceptable consequences for the mistakes. Perhaps society getting used to the sorts of things it will fail at and reducing how much time we try to make the LLMs play in that 70% wrong sort of use case.

    I see LLMs as replacing first line support, maybe escalating to a human when actual stakes arise for a call (issuing warranty replacement, usage scenario that actually has serious consequences, customer demanding the human escalation after recognizing they are falling through the AI cracks without the AI figuring out to escalate). I expect to rarely ever see “stock photography” used again. I expect animation to employ AI at least for backgrounds like “generic forest that no one is going to actively look like, but it must be plausibly forest”. I expect it to augment software developers, but not able to enable a generic manager to code up whatever he might imagine. The commonality in all these is that they live in the mind numbing sorts of things current LLM can get right and/or a high tolerance for mistakes with ample opportunity for humans to intervene before the mistakes inflict much cost.