There are constraints, real world constraints to this stuff. It’s just that the consequences manifest themselves in 20+ years. Whoever thinks that in our current position it is a good idea to burn the equivalent of a nations energy consumption on producing fake pictures of cats and junior level code that is riddled with hard to find mistakes is just nuts. There are sensible applications, but going gonads to the wall crazy in search of a problem just because people have a solution that attracts investor money is obscene to say the least.
The ecologic consequences of this AI craze will be catastrophic.
Yes, but those models get continuously retrained, made bigger, used with less context restraints. Any CS undergrad knows that quadratic problems don’t scale, yet here we are reactivating decommissioned NPPs on, well, a whim because there isn’t even a business plan.
And I think? that is my chief concern, how the algorithms interpret that data. The data is either based on - information silo’s ‘owned’ by Google/Meta/X/Tic Tok, etc (no bias there ) or scraped from the Web with as much respect for privacy and IP as the CCP, which gives me an ethical dilemma.
And then you factor in the environmental concerns. Obscene power consumption and a wasteland for future generations… but hey ‘Drill baby, drill’!
Not sure if the AI tech is the problem, or the hype going on with some companies thinking that accuracy scales per watt used to train the models.
It seems that nowadays the model size is marketing term to induce hype on stock market. Once the initial hype cools down, the price of energy will once again become relevant.
You are not off mark. But there‘s also a lot of walled gardens out there. Who has the most valuable sets of data? How can it now be processed in a way that did not exist before?
It‘s just a new tool in the shed. Happens all the time. Also it‘s overhyped, because „AI“ isn‘t even there yet.
We got GenAI and Agents (=GenAI asking questions to itself for a more refined result).
The most fun part for me is how you can stop treating your computer like a computer and start treating it like a human. We already know how to do that. Just talk to it and reflect on every reply. Don‘t trust blindly. Nothing new.
Prompt engineering my ass! It‘s called „having a conversation“.
There was a recent political event in the US where all the holders of front row seats were tech oligarchs. No manufacturers. No philosophers or thought leaders. Just tech. That tells us all we need to know. Data is the coin of the realm. As for AI, it is a tool for making the wealthy ever wealthier while disguised as a productivity tool for everyone.
I am at the part in the series of sci-fi books popularly called The Expanse where the Belters dropped rocks on Earth, killing at least 1/3 of the population of 30b. Videos like this make me think that it’s maybe time for someone to drop those rocks. Humans are redundant.
I‘m thinking my next Quarterly Business Review meeting will be summarized in a rap video…
Trackin’ metrics, that’s how we roll Key Performance Indicators, gonna reach our goal Revenue climbin’, efficiency tight Measuring success with data so bright
The city administration in the city where I live, recently published a new proposition to centralize schools and consequently close several schools in the city. The reason for this was declining birth rates and that it would create a better learning environment for the kids. This proposition was said to be based on the latest research on the subject.
Today the bomb dropped…
The proposition was probably AI generated. Most references to research pointed to books and articles that doesn’t exist.
The authors of the fictitious books are real and are renowned for their research in the area…but they never said or published what they were quoted for.
“I have been misunderstood and misquoted before, but never quoted on things I’ve never said!” said one of the authors.
) or scraped from the Web with as much respect for privacy and IP as the CCP, which gives me an ethical dilemma.
