walt74
Joined 14 years ago
Karma
26
Submissions
88
Member since
Sep 30
Recent submissions
Comment · 5 months ago
Comment · 2 years ago
Ofcourse this article rubs the open source dogma believers here the wrong way and i'll get downvoted to hell, but i wholeheartedly agree with the author, and i think that you guys don't get what AI-systems are, at their basic level.
The are information interpolators, and all the information you can interpolate from the training data is readily present in latent space, waiting to be discovered by a prompt.
There is this argument: But but you can find a bomb instruction in a chemistry textbook. No, you can't. Without checking this, but i think you'll have a hard time finding any chemistry textbook that explicitly gives you a bomb instruction. Ofcourse, yes, you can find all the information necessary to build that bomb in that text book, but the key difference to a latent space full of interpolated data points is that: You have to sit down, find that information that is scattered throughout that textbook, write it down, interpolate that knowlegde yourself, write that down, and then you have a bomb instruction -- except you'll have written it for yourself.
Not so with latent spaces. The bomb instruction is already there, interpolated from all the data points, just waiting to be prompted, and that is easy peazy with, yes, open source models.
So spare me the whining about anachronistic software dev dogmas from the 90s and arrive in the present, pretty please.
Comment · 2 years ago
Agreed, but to tackle the problem from that perspective would require making LLMs a public good, preferably run by the state, akin to public libraries. This could not only solve for the copyright problem, the state may even make it mandatory for publishers to contribute their published writings to the public LLMs. I'm sure libertarian tech bros have that in mind when they insist on open source development (which then opens another whole can of worms when you consider interpolative knowledge as intellectual nuclear fission, but that's another story).
Comment · 3 years ago
1. Przybylski again. He continues to generalize from studies focussing on screentime that everything digital is harmless, including social media. I don't trust his work.
2. [Children's brains shaped by their time on tech devices, review shows](https://medicalxpress.com/news/2023-11-children-brains-tech-...) and [Does screen use really impact thinking skills? Recent analysis suggests it could](https://medicalxpress.com/news/2023-11-screen-impact-skills-...)
> The evidence review, published today in Early Education and Development, is an analysis of 33 studies which use neuroimaging technology to measure the impact of digital technology on the brains of children under the age of 12. In total, more than 30,000 participants are included.
> In particular, the research finds screen time leads to changes in the pre-frontal cortex of the brain, which is the base of executive functions such as working memory and the ability to plan or to respond flexibly to situations. It also finds impacts on the parietal lobe, which helps us to process touch, pressure, heat, cold, and pain; the temporal lobe, which is important for memory, hearing and language; and the occipital lobe, which helps us to interpret visual information.
> "It should be recognized by both educators and caregivers that children's cognitive development may be influenced by their digital experiences," says the study's corresponding author, Chair Professor Hui Li, from the Faculty of Education and Human Development Faculty of Education and Human Development, at The Education University of Hong Kong.
It traces the history of fuzzy archives back to Aby Warburgs Mnemosyne Atlas and Borges and goes on to explain the various effects on learning, including chances and risks, and dissects various points of critique from delusions to parrots, some of which prevail, many of which vanish, once you strip AI from cognitive woo.
This is the whole piece (14k words no less): https://goodinternet.substack.com/p/on-interpolatable-archiv... and it contains links to its 4 parts, if you like to read in smaller chunks.
It was quite a lot of work over some weeks, and I hope you guys appreciate. Feedback welcome.