Jan Penfrat"Asking enslaved people if they actually want to work the plantations for free would kill the cotton industry."
-- Nick Clegg
Martin Rundkvist
dotmavriq@mrundkvist @ilumium
I sympathize with your sentiment across all of your posts critiquing "AI" per se...but I don't think we can shy away from the strides done with the technology, like strides made in predicting cancer behavior, breakthroughs in the medical field et. c.
And as a programmer... well... the numbers seem to be in and are mostly cynical in regards to whether or not an actual improvement has occured across the board from a productivity perspective.
But I can tell you confidently that utilizing AI to do my work or contribute to open source projects feels fantastic as I sit with models and get a hang of "what type of weight they can pull" and from there I gauge what I can get out of it and there is A LOT of cognitive load that I can instead focus on Software architecture rather than sifting through documentation for a lot of mundane/technically redundant programming.
Is it currently enshittifying the world? Hell yes.
Did it bite the job market and cause me pains in 2024 and the beginning of 2025? Yep. You bet.
Is it all bad? No. And we can't understate the improvement vector. How many times did you hear us b*tch and moan over water and electricity usage for Google searches as a society?
Right? ... well they're inching in on having regular AI-query at that level of energy/water consumption and the trajectory for it is fantastic.
It's easy to get lost in "what generative AI has done to social media and academic writing" or in my case "what AI has done to programming as a whole" ...and I'm right there with you. We just can't write off the entire thing. I'm not Kurzweil levels of "bracing for AGI" or any of that (the name is horrible, we are not even emulating human cognition per se), we haven't seen any proper signs, I remain skeptic... but even LLMs as such ARE doing a lot of interesting things in the right places... we just have this inflated repackaging of it right now as it's the only thing interesting that has happened in Tech since "Moores Law for hardware, graphics and network speeds" started to cool down.
Rant over. Send me rårakor
I sympathize with your sentiment across all of your posts critiquing "AI" per se...but I don't think we can shy away from the strides done with the technology, like strides made in predicting cancer behavior, breakthroughs in the medical field et. c.
And as a programmer... well... the numbers seem to be in and are mostly cynical in regards to whether or not an actual improvement has occured across the board from a productivity perspective.
But I can tell you confidently that utilizing AI to do my work or contribute to open source projects feels fantastic as I sit with models and get a hang of "what type of weight they can pull" and from there I gauge what I can get out of it and there is A LOT of cognitive load that I can instead focus on Software architecture rather than sifting through documentation for a lot of mundane/technically redundant programming.
Is it currently enshittifying the world? Hell yes.
Did it bite the job market and cause me pains in 2024 and the beginning of 2025? Yep. You bet.
Is it all bad? No. And we can't understate the improvement vector. How many times did you hear us b*tch and moan over water and electricity usage for Google searches as a society?
Right? ... well they're inching in on having regular AI-query at that level of energy/water consumption and the trajectory for it is fantastic.
It's easy to get lost in "what generative AI has done to social media and academic writing" or in my case "what AI has done to programming as a whole" ...and I'm right there with you. We just can't write off the entire thing. I'm not Kurzweil levels of "bracing for AGI" or any of that (the name is horrible, we are not even emulating human cognition per se), we haven't seen any proper signs, I remain skeptic... but even LLMs as such ARE doing a lot of interesting things in the right places... we just have this inflated repackaging of it right now as it's the only thing interesting that has happened in Tech since "Moores Law for hardware, graphics and network speeds" started to cool down.
Rant over. Send me rårakor
@dotmavriq
There are many different machine learning techniques. I am a warm friend of the analytical ones where you e.g. show the machine laser altimetry models of 25 known charcoal production sites and then ask it to identify similar ones in a huge dataset. I am opposed to the use of LLMs to generate prose and answer factual questions. Because there is no intelligence there. It's autocomplete.
@mrundkvist
Don't you think the current iteration will burst veeeery very soon though.
As in, consumers/lawmakers will get wise to exactly how inaccurate and flawed they are and try to choke them out of societal importance.
Don't you think the current iteration will burst veeeery very soon though.
As in, consumers/lawmakers will get wise to exactly how inaccurate and flawed they are and try to choke them out of societal importance.
@dotmavriq
The venture capitalists will soon realise that there is no profit to be made from investments in the technology.
The venture capitalists will soon realise that there is no profit to be made from investments in the technology.