It’s crazy. Literally the only valid thing you can say about LLM use is “in the hands of an experienced person LLMs can be a force multiplier”.
But no, we have to replace entire companies with it. All the problems of LLMs stem from inexperienced people using it (by inexperienced I mean not skilled in the domain in which the LLM is being applied).
I think of inexperienced people using power tools. Power tools theoretically make your life easier, but beginners will strip screws or snap bolts right and left. Or with ikea furniture the screw threads in the "wood" get pulled right out.
With screws or LLMs, craftsmen can speed up and do the finer details by hand.
I will argue that we are going forward - for better or for worse. In my humble opinion, taking into account the sad state of the world today, any change is welcomed.
The world will get a lot crazier in the years ahead, on the path to superhuman intelligence. Disruptions caused by past inventions, like the printing press, the steam engine, electricity or the Internet, will look unremarkable by comparison.
Hank Green has had the best take I've heard which is AI is definitely something but everyones still figuring out what it is. It will likely end up being nothing like it is today.
The commencement/graduating speeches including AI in the video at the end are infuriating. The message from people giving those speeches is literally "deal with it" and how tone deaf do they have to be to patronize those booing students with "I understand your fear".
The graduation episode where the AI readout missing some student names and then the college saying "we used AI to readout and some names were missed. We will not redo and you will not see your name on stage" is the worst.
I believe the main value of AI comes not from its productivity gains but because AI will increasingly become a tool for evading responsibility and accountability for actions in economic, social and worse even military functions.
I was surprised to find out how much hate there is for AI in art.
An artist, Yuumei, is the perfect candidate to use AI– drawing by hand since early 2000s, wrist injury precluding heavy work.
People seem to think art should be done only by humans, that AI steals art, and is bad for the environment.
But she wants to use it to be able to produce the work she wants, including comics with lots of art and such. Given that she's ultimately still responsible for the creative direction and result, this seems like something AI is greatly help for.
> tools that let users build personalized mini apps tailored to their needs.
We’ve reinvented Java Applets. I mean, I do like the idea behind that sort of stuff, it’s just that all sorts of little things break along the way. For example, I asked Claude to put together a specific recipe, it could do that, I got my Artifact/cooking widget/whatever. It even let me switch between metric and imperial (and didn’t save that preference) and let me change the quantity and updated the ingredient amounts (except the phone going to sleep led to it all resetting).
Sometimes I feel like we are very much stuck in being able to produce things but they simply aren’t high enough quality (which might take years or decades more of model training and efficiency improvements) and also that maybe we’re doing things a decade too soon. Imagine trying to build AI data centres with 2010 or 2000 hardware and how limited the models you’d be able to run would be. Maybe that’s also why the current outcomes are sometimes shitty. The other theory is that there’s simply not enough high quality data to train truly good models and we’ll plateau and model collapse in training will be common.
LLMs as a technology are neutral and largely act as amplifiers.
Now, there is not an ounce of decency between our SV overlords and I have zero trust they will choose to amplify the right things. On the contrary, their apparent ideal state is a vast swath of technoserfs force fed ever more content and ads and more content, getting by on Uber-for-everything, where you spend what little money you have saved up being a delivery boy on things you don't own. We will stumble about in barren apartments, living a fake life through VR goggles, watering virtual plants with virtual water, all of which we pay for of course. All the while Zuckerberg, Musk, and Thiel are tucked away on the moon, their vile hands clinging to a last hope of immortality, just as hopeless as the people below.
Contrary to the chorus, I find this mad episode in our history has been going surprisingly well. I mean, it's a lot to process and the speed is ludicrous. What did we expect? What is the baseline? 80% approval in face of unprecedented uncertainty?
We are adjusting, we talk about things. I hope we will keep doing that.
Technology is a side effect and distraction from the real problems of matter and the ones which matter. Democracy is not a guarantee of making the right choices nor is theocracy nor is any form of government. We are driven blindly ever forward by our nature to survive the following common factors: hunger, thirst and exposure.
AI companies keep repeating that AI will replace white-collar work, and that what they're building is a potential doomsday device. How did they expect graduates to react? Cheer them on? Feel hopeful?
abstract thoughtslop I've been thinking about about this:
Because most people predicting the future don't like whats coming, and there's not many people interesting in actual predicting, I think the act of having a good idea of what's next makes the future less known.
(As I understand it), on the stock market, making a better day-trader bot makes the short-term squiggles of the price more random. Generalizing, if you know with really good odds what's going to happen a few months from now, and want to profit from it, I think there might be a similar effect. I'm not talking about insider trading or knowing which policies are going to be enacted, I mean just people who just forecast better with public info, that the act of better knowing what's going to happen makes the world more complex. (coordination issue?) I also think if like 70+% of the population was better/motivated at/to forecast stuff we would be better off.
19 comments
[ 2.5 ms ] story [ 30.1 ms ] threadBut no, we have to replace entire companies with it. All the problems of LLMs stem from inexperienced people using it (by inexperienced I mean not skilled in the domain in which the LLM is being applied).
I think of inexperienced people using power tools. Power tools theoretically make your life easier, but beginners will strip screws or snap bolts right and left. Or with ikea furniture the screw threads in the "wood" get pulled right out.
With screws or LLMs, craftsmen can speed up and do the finer details by hand.
The graduation episode where the AI readout missing some student names and then the college saying "we used AI to readout and some names were missed. We will not redo and you will not see your name on stage" is the worst.
I believe the main value of AI comes not from its productivity gains but because AI will increasingly become a tool for evading responsibility and accountability for actions in economic, social and worse even military functions.
An artist, Yuumei, is the perfect candidate to use AI– drawing by hand since early 2000s, wrist injury precluding heavy work.
People seem to think art should be done only by humans, that AI steals art, and is bad for the environment.
But she wants to use it to be able to produce the work she wants, including comics with lots of art and such. Given that she's ultimately still responsible for the creative direction and result, this seems like something AI is greatly help for.
Example hate video and comments: https://www.youtube.com/watch?v=495VOuAnCJM
We’ve reinvented Java Applets. I mean, I do like the idea behind that sort of stuff, it’s just that all sorts of little things break along the way. For example, I asked Claude to put together a specific recipe, it could do that, I got my Artifact/cooking widget/whatever. It even let me switch between metric and imperial (and didn’t save that preference) and let me change the quantity and updated the ingredient amounts (except the phone going to sleep led to it all resetting).
Sometimes I feel like we are very much stuck in being able to produce things but they simply aren’t high enough quality (which might take years or decades more of model training and efficiency improvements) and also that maybe we’re doing things a decade too soon. Imagine trying to build AI data centres with 2010 or 2000 hardware and how limited the models you’d be able to run would be. Maybe that’s also why the current outcomes are sometimes shitty. The other theory is that there’s simply not enough high quality data to train truly good models and we’ll plateau and model collapse in training will be common.
cannot be outsourced to someone prompting an AI / LLM / whatever the next technology is / from Guntur or Wajir.
Very few people are irreplaceable.
https://www.fastcompany.com/91544842/ai-slop-facebook-conten...
Now, there is not an ounce of decency between our SV overlords and I have zero trust they will choose to amplify the right things. On the contrary, their apparent ideal state is a vast swath of technoserfs force fed ever more content and ads and more content, getting by on Uber-for-everything, where you spend what little money you have saved up being a delivery boy on things you don't own. We will stumble about in barren apartments, living a fake life through VR goggles, watering virtual plants with virtual water, all of which we pay for of course. All the while Zuckerberg, Musk, and Thiel are tucked away on the moon, their vile hands clinging to a last hope of immortality, just as hopeless as the people below.
We are adjusting, we talk about things. I hope we will keep doing that.
Because most people predicting the future don't like whats coming, and there's not many people interesting in actual predicting, I think the act of having a good idea of what's next makes the future less known.
(As I understand it), on the stock market, making a better day-trader bot makes the short-term squiggles of the price more random. Generalizing, if you know with really good odds what's going to happen a few months from now, and want to profit from it, I think there might be a similar effect. I'm not talking about insider trading or knowing which policies are going to be enacted, I mean just people who just forecast better with public info, that the act of better knowing what's going to happen makes the world more complex. (coordination issue?) I also think if like 70+% of the population was better/motivated at/to forecast stuff we would be better off.