Page 5

^What are you suggesting? I know what the letters stand for, if that's what you're referring to

2/26/2026 4:51:53 PM

But are you black?

2/26/2026 5:57:02 PM

I have definitely said FAQ phonetically back in the day

[Edited on February 26, 2026 at 7:16 PM. Reason : lemme go find an animated Gee Eye Eff file to post]

2/26/2026 7:16:13 PM

2/26/2026 8:35:59 PM

I'll give plus points to anthropic for an act of resistance

2/27/2026 5:17:10 PM

Quote :

"Anthropic’s contract with the Pentagon is worth as much as $200 million. The company had sought to have protections written into the deal against the use of its models for autonomous weapons and mass domestic surveillance."

I don't get the big deal about the bolded part. Like as much as the NSA spies on us, collecting phone records, metadata, etc...what could AI possibly do thats not already done now?

2/27/2026 9:08:27 PM

ask AI

2/27/2026 10:24:36 PM

^^The concern about AI and mass surveillance isn’t just that governments collect large amounts of data — agencies like the National Security Agency have done that for years. The key difference is what AI allows them to do with that data.

In the past, massive data collection still required large teams of human analysts to interpret it. That created natural limits. AI removes those bottlenecks. It can automatically analyze millions of conversations, detect patterns, infer political leanings or emotional tone, map social networks, and flag “suspicious” behavior in real time. Instead of just recording what people did, AI can model and predict what they might do.

That shift—from passive storage to automated interpretation and prediction at population scale—is what raises civil liberties concerns. It lowers the cost and friction of monitoring everyone, making broad, always-on surveillance technically feasible.

Companies like Anthropic worry that their models could enable automated repression or large-scale behavioral scoring systems, similar to concerns often raised about systems in China. The debate isn’t about whether surveillance exists today—it’s about how dramatically AI could expand its power and reach tomorrow.





I asked(in around 100 words) "whats the problem with always-on surveillance because if anything it seems like a good thing...it deters crime, terrorism, etc

Quote :

"Supporters of broad, always-on surveillance argue it increases safety, deters crime, and helps prevent threats before they happen. If someone isn’t doing anything wrong, they believe there’s little to fear. Critics argue the problem isn’t just spying — it’s power imbalance and long-term risk. When authorities can monitor everyone, people may self-censor, reducing genuine freedom. AI systems can also make errors, falsely flagging individuals at scale. Surveillance programs often expand beyond their original purpose, and stored data can later be reinterpreted under new laws or leadership. The core debate is whether increased security is worth sacrificing private, unmonitored space."

I'm a strong proponent of "If someone isn’t doing anything wrong, they believe there’s little to fear."

2/27/2026 10:40:24 PM

^ Do you want AI mistaking your account for terrorist activity and stealing your life savings?

Think SWATing instances where the police barge into the wrong home and blow some dude away because they got the address wrong. Do you want a model that hallucinates responsible for determining if you get detained for 12 hours? Even if you have nothing to hide they can freeze your accounts by mistake or deny all kinds of services that you may use and need. Including banking and credit. There have been countless cases of people going to buy cars with cash and getting it swiped by the government or police at a stop. Any interaction with gov has the risk of a mistake happening.

Usually when the gov fucks up massively it becomes your responsibility to prove it. If you aren't a pretty girl or some aggrieved class the chances of you winning or getting anyone to care are low.

Say the mistake rate for targeting the wrong person is 0.5% and with current methods they wrongly accuse and harass 5000 people. With new AI tools they cover say 20x as many people. Assuming the mistake rate is the same that's an extra 95,000 people that are going to be wrongly harassed.





[Edited on February 28, 2026 at 7:32 AM. Reason : It's estimated between 1 and 5% of convictions are wrongful]

2/28/2026 7:20:56 AM

An extra 95000 is like chump change. We used to do magnitudes that amount each year at the border

2/28/2026 4:55:31 PM

https://newrepublic.com/post/207693/palantir-ceo-karp-disrupting-democratic-power

Quote :

"Palantir CEO Alex Karp thinks his AI technology will lessen the power of “highly educated, often female voters, who vote mostly Democrat” while increasing the power of working-class men."

3/12/2026 8:37:32 PM

https://www.404media.co/the-removed-doge-deposition-videos-have-already-been-backed-up-across-the-internet/

3/15/2026 8:09:50 AM

https://www.mprnews.org/story/2026/03/18/fargo-polices-use-of-ai-raises-questions-after-suspect-says-facial-recognition-failed

3/20/2026 9:52:03 AM

One slop factory down

https://www.hollywoodreporter.com/business/digital/openai-shutting-down-sora-ai-video-app-1236546187/

3/24/2026 7:22:15 PM

Quote :

"What you made with Sora mattered"

LOL! Sure, it did! . . .

3/24/2026 8:46:02 PM

Quote :

"I agree 100% with Steven Vickers that authors will be falsely accused of using AI in the coming years. In particular, authors like myself whose works were stolen to train AI systems are at risk. When we write in our own unique styles, AI trained on us will quite likely flag us as using AI."

https://bsky.app/profile/jasonsanford.bsky.social/post/3mhuw4onfhk24


[Edited on March 26, 2026 at 7:50 PM. Reason : E]

3/26/2026 7:50:01 PM

I always knew Mary Shelley was a con artist. Now it finally makes sense!

3/26/2026 8:02:14 PM

Pro ai camp is losing the hearts and minds. Some state legislator had his house shot up for supporting Data centers. Public sentiment is broadly against ai.

I think the only solution is for heavy handed regulation so the public feels confident we have control. One easy thing is a windfall tax or some other mechanism to force tech companies to shoulder basically all of the Costs of infra build outs. No local tax breaks. 2Nd thing is all models have some level of free access— these models are trained on our data and we should have access to their capabilities. 3Rd thing is something like irelands creativity fellowships— basic income for anyone who wants to pursue a passion.

4/7/2026 3:19:14 PM

the public doesn't care about having AI "under control" they care about electricity prices going up (which started after covid and hyperscalers have already started saying they're going to pay for (their own in some cases) the electrical infrastructure) and are tired of people using AI to make slop (even though they'll happily consume it anyways). giving in to violent psychos who don't even understand what they're protesting against is the last thing we should be doing.

[Edited on April 7, 2026 at 8:43 PM. Reason : public sentiment in China is very pro AI. asking ourselves why that is is an important exercise. most of the negative sentiment in the west is misplaced grievance-seeking]

[Edited on April 7, 2026 at 8:43 PM. Reason : anti-datacenter is today's anti-5G tower]

4/7/2026 8:42:45 PM

China: famous lovers of the arts

4/7/2026 10:23:37 PM

Was anti-5g tower thing a real thing?

4/7/2026 10:48:12 PM

no, it wasn’t

data centers polluting low-income, mostly black communities is very real, though

4/7/2026 10:50:38 PM

^ We should annex Berkley Cafe and put it there. Los Angeles has a ton of large data centers in high rises.

https://en.wikipedia.org/wiki/One_Wilshire The entire block around it is nothing but data center buildings in prime downtown real estate. Raleigh could lead the way and protect the poors!

4/8/2026 10:15:52 AM

Quote :

" From a survey of 1,572 people aged 14-29: • Hopefulness regarding AI has dropped from 27% to 18% over the past year, while excitement fell from 36% to 22%. • Anger toward AI increased from 22% to 31%, while anxiety remains flat affecting 42% of Gen Z. "

https://www.nytimes.com/2026/04/09/style/gen-z-ai-gallup-study.html

4/9/2026 12:48:04 PM

Quote :

"data centers polluting low-income, mostly black communities is very real, though"

Yeah, but if it doesn't affect me, it must be made up!

4/9/2026 7:45:40 PM

https://blog.samaltman.com/2279512

4/10/2026 8:35:07 PM

Quote :

"The world deserves huge amounts of AI"

QFT, but I don't think we mean it in the same way. At all.

4/10/2026 9:17:17 PM

Quote :

"Words have power too. There was an incendiary article about me a few days ago. Someone said to me yesterday they thought it was coming at a time of great anxiety about AI and that it made things more dangerous for me. I brushed it aside."

So is he going to link any negative article about him to his security?

4/10/2026 10:28:31 PM

https://bsky.app/profile/andymasley.bsky.social/post/3mjcw4loigc2s

4/12/2026 2:30:52 PM

Now do one about using it instead of just training it.

4/12/2026 2:58:10 PM

Not as crazy as training. Estimates are 0.3 to 3 Wh per query. Roughly like running a lightbulb for 10 minutes.

Let's say half the internet users use AI. 125 million people. Doing 5 queries per day.

225 million Kwh per year. Reasonable guess 100 to 500 million tweaking the above. Using a kWh to tones co2. something like 56000 tons. Way less than coke cans in a day for a full year of use. Paid for by banning the poors from buying coke on snap!

[Edited on April 12, 2026 at 3:24 PM. Reason : A]

4/12/2026 3:19:36 PM

Marginal resource usage per generated token is a continuously dropping metric. Hard to measure because it can vary drastically depending on the model. But it’s relatively small no matter how you slice it. But the usage of ai is very high so it’s a big impact— if demand is high for something though and you cut off the supply you increase cost and decrease access. I don’t want to live in a world where only upper class people have access to ai. Cutting off the data center buildouts is the wrong move but we do need to make sure they’re following environmental regulations though and I would want to see a requirement for 100% green energy for any new ai datacenter— paid for by the tech companies building it.

4/12/2026 4:09:43 PM

Seems like we're a LONG way from that. . .

4/12/2026 6:10:13 PM

https://www.reddit.com/r/singularity/comments/1sjtebt/sam_altmans_home_targeted_in_second_attack/

4/12/2026 8:58:01 PM