Yeah yeah, AI data centers use so little water, all AI doomers are bonkers, blah blah blah. https://www.1011now.com/2026/09/30/more-questions-than-answers-about-lincolns-google-data-center-water-electricity-usage/
They say they are scared because of the recent incidents like the HuggingFace hack, but come on, you just need competent IT staff to setup a firewall to avoid that 🤷♂️.
I think the reason for them wanting to slow down is other. Maybe they are running out of money for training runs. Or maybe they want to avoid investor pressure because of unfulfilled promises.
Whatever the reason is, just do not trust them, they are compulsive liars.
(2/2)
Altman and Amodei have repeatedly said you have to rush AI research above everything else because:
➡️ it will lead to curing cancer, solving climate crisis, exceptional wealth for everyone, 4 work day weeks...
➡️ It will be a disaster if China wins the race.
They pushed hard to avoid regulations in order to move fast, and have reduced the safety teams to a skeleton crew.
And suddenly both of them say they should better slow down.
1/2
I have only seen the first 15 minutes of the new video by #GamersNexus ("216,000,000 Spy TVs | The LG Smart TV Problem") and oh boy this shit is scary. Watch it right now: https://www.youtube.com/watch?v=6IFVTcM28KA
I'm releasing my #OpenSourceHardware #SuperNintendo cartridge and programmer!
➡️ XiROM Cartridge: https://gitlab.com/doragasu/xirom-cart
➡️ XiROM Programmer: https://gitlab.com/doragasu/xirom-prog
The cartridge features are:
➡️ Support for HiROM (up to 32 Megabit) and ExHiROM (up to 63.5 Megabit)
➡️ Support for FastROM
➡️ Multi region SuperCIC implementation
The XiROM Programmer can, of course, program XiROM carts, including the SuperCIC firmware. It can also dump SNES standard cartridges.
#OpenSource #SNES #SNESDev
Big companies putting a cap on #GenerativeAI token spend is just them admitting the thing does not work, but they do not want to admit the thing does not work.
If the slop machine worked to increase productivity like they have been saying, they would continue the initial discourse of "maximizing token spend = maximizing productivity".
The reality is productivity does not increase. In fact if your definition of productivity takes into account tools cost, it abruptly plummets.
Even Macs look like reasonably priced computers now!
Who would have told us that in the current market situation, we can say that "The #Switch2 is a competitively priced console".
My little SNES cartridge programmer/dumper.
#FreeCAD #FreeCADFriday
Tech CEOs promised #GenerativeAI would make our life easier. Let's see what they gave us so far:
* Nice but expensive tools for software vulnerability research and reverse engineering
* Hallucinations
* AI psychosis
* Massive technical debt
* People cognitively exhausted having to babysit AIs
* Everything more expensive, specially anything using a considerable amount of RAM and storage
* Pollutant datacenters potentially near you
* Climate change acceleration
* Economical collapse acceleration
😬
Diodes + Capacitors = ?
Anyone trusting #Microsoft to keep secure any data more sensitive than a grocery shopping list, is so completely delusional. They have a backdoor in #Bitlocker that just needs to deploy a few files on a pendrive or on the EFI partition if USB access is not available 🤦♂️. Ridiculous.
➡️ https://www.techspot.com/news/112410-security-researcher-microsoft-secretly-built-backdoor-bitlocker-releases.html
➡️ https://github.com/Nightmare-Eclipse/YellowKey/tree/main
#Microslop #Security #Backdoor
It's always interesting seeing how in many games, #Linux can crush #Windows 11 perf. It's even more impressive considering games are Windows native and on Linux they run on top of a compatibility layer. The performance difference can be specially high for AMD GPUs with 8GB VRAM or less. https://www.youtube.com/watch?v=pfRMGQzhIX0
This has become an obsession. Played the whole thing thrice, last time in hardcore mode, and the more I play, the more impressed I am.
The many branches you can take along with the trait system make this the VN with the best replayability I have seen by far, on each playthrough you are surprised on completely unexpected ways, and continue digging into the wonderfully written story. Too bad I don't know how much I'll have to wait for the finale.
#ScarletHollow
I've been not having too much sleep this week... because I purchased #ScarletHollow and I have been completely absorbed. Yesterday finished chapter 5 and the final chapter 6 has not yet been released. But the branching depth looks so deep, I'm starting another playthrough with a different build and different choices, I am intrigued by many things I could have done differently... Awesome game. If the conclusion does not disappoint, it will be one of my all time favorites. https://www.youtube.com/watch?v=LwZ9wlVV42U
Wow, #Copilot editing Pull Requests to insert ads is something horrible in so many levels. #Microsoft is really determined to kill #GitHub. https://notes.zachmanson.com/copilot-edited-an-ad-into-my-pr/
➡️ #SamAltman: With enough computing power, #AI can cure cancer.
➡️ Reality: We were wasting $15 million a day on #Sora inference costs to generate brainrot and memes. And let's not talk about training.
https://www.revolutioninai.com/2026/03/%20chatgpt-gpt-54-mini-silent-switch-march-2026.html
#OpenAI #GenerativeAI #GenAI
➡️ #JensenHuang: "If that $500,000 engineer did not consume at least $250,000 worth of [#GenerativeAI] tokens, I am going to be deeply alarmed"
➡️ Reality: A friend of mine working for a software company that forced #GenAI usage about one or two years ago, told me they have been asked by management to try not spending more than $500/month on the #slop machine.
Now think what will happen when the tech stops being subsidized.
https://www.businessinsider.com/jensen-huang-500k-engineers-250k-ai-tokens-nvidia-compute-2026-3
#LLM #Claude
This is what the slop machine is leading software development to: junior/middle devs generating slop at an astounding rate, and senior devs completely overloaded with code reviews, to the point they cannot thoroughly review them because the slop pile never ends.
Also junior devs that never will stop being junior because they'll learn nothing from the process.
Windows was never that stable, but since Microsoft stated "30% of our code is written by AI", it's directly a shitshow. What I didn't see coming is AWS seeing its stability seriously compromised because of #LLM generated code. https://arstechnica.com/ai/2026/03/after-outages-amazon-to-make-senior-engineers-sign-off-on-ai-assisted-changes/
#GenerativeAI