Deval Patrick's Wins (@DevalsWins) header
Deval Patrick's Wins (@DevalsWins) profile picture

Celebrating the wins and victories of the 7th Governor of Massachusetts, Deval Patrick! Not affiliated with Fmr Governor Patrick. Let's go!

Joined July 2024
Photo by Deval Patrick's Wins (@DevalsWins)Photo by Deval Patrick's Wins (@DevalsWins)Photo by Deval Patrick's Wins (@DevalsWins)Photo by Deval Patrick's Wins (@DevalsWins)Photo by Deval Patrick's Wins (@DevalsWins)Photo by Deval Patrick's Wins (@DevalsWins)Photo by Deval Patrick's Wins (@DevalsWins)Photo by Deval Patrick's Wins (@DevalsWins)Photo by Deval Patrick's Wins (@DevalsWins)Photo by Deval Patrick's Wins (@DevalsWins)
"guys we just lost. Clearly we just have to get more racist"
It’s time to start saying openly “racist” things. You either fight for your people on the basis of race or you watch your children be subdued by third world mongrels. There is no other option.
5
157
Deval Patrick's Wins retweeted
Haven’t heard about the parliamentarian in a while excited to see that guy again
9
278
5
3,261
33,216
Deval Patrick's Wins retweeted
Gimme three months and his golf buddies in Florida wouldn’t even vote for him. I’ll smoke him by 7. Bet on it.
3,336
40,909
2,958
352,361
5,795,425
one thing to watch out for tonight is the number of votes each candidate in a race is getting that should tell you which candidate is going to win that election
1
3
119
first day of standing lessons
Met with Venezuelan Foreign Minister @oliverblanco today. Reconciliation and institutional reform are Venezuela’s path to an enduring republic. The United States will keep supporting this Venezuelan-led process, which is laying the groundwork for free, fair, and credible elections.
1
118
Deval Patrick's Wins retweeted
“I joined the military to serve Israel, now I want to join the US Senate to serve Israel.” - Mike Rogers
414
1,468
308
5,577
642,958
"I don't think you understand, what happens in Israel matters to Jews everywhere!" "So you have dual loyalties to Israel and the United States?" "OMG ANTISEMITE HOW COULD YOU SAY THAT"
I really don’t think people understand: the Jewish people is ONE PEOPLE— wherever we live on earth. Israel is home to ½ of the Jewish people globally. So, what happens in Israel matters deeply to Jews everywhere. Anything that is a threat to Israel is a threat to Jews everywhere.
1
213
NEW Final Marist (A+) Poll: (October 2 - October 11) 🔵 James Talarico (D) 🔴 Ken Paxton (R) 🔵 Gina Hinojosa (D) 🔴 Greg Abbott (R)
1
10
880
Marist leak is crazy spamming the buy on Paxton rn lmao
1
330
These guys always complain that someone yucky is showing the "content" in order to "defame India" but never question why the content exists in the first place
Replying to @aravind
Sir, most of this drivel comes from Dalit handles only. They are the ones who amplify and provide content to foreigners.
1
3
109
total cultural kirktory right wingers making kirk jokes
Sources tell me James Talarico is stable. Please god
3
123
they're jerking off to fanfiction 😭
Replying to @CommissionerPW
The rumors about Talarico have been WILD. The most credible one I have heard is that something big came to light, and he's being given a choice: disappear on your own terms or they are going public. I think Talarico's campaign is trying to buy someone off, reach a private settlement or figure out a way to spin things. I've been waiting for this "October Surprise" for months. From what I understand, a previous student/intern/ whatever (most likely a male) has come forward and said that Talarico molested/raped/ groomed him when he was a child.
2
65
Waiting at gate at Tel Aviv airport to go back to US, checked my email. A reader of my Substack wrote this below. It's terrifying to me, but we have to know this is real, so we can fight it. I'm a 59-year-old Gen Xer; v. few people my age & older have any idea this is going on.
4
200
Replying to @matthew_petti
Kind of glib to point out but remember when it was beyond the pale anti-semitism to even suggest that zionists might have dual loyalties
12
12
318
4,975
Deval Patrick's Wins retweeted
Tens of thousands of goyim got one shotted into thinking an LLM is le heckin based and christpilled when it was proven it just picks whoever is in the center of the pic. It can only go downhill from here.
Replying to @BaT_MaN_23
Done.
32
41
2
669
19,711
Deval Patrick's Wins retweeted
have model welfare bros considered that there are tremendously more insects than there are language models
171
30
29
1,156
97,426
moron: "Claude, serious question" claude: "I am taking the question seriously" moron: "oh my God" be sure to ask the ai it's pronouns otherwise you might give a GPU gender dysphoria!!
Good morning to me! Last time I had already justified it. Deepseek only rewrote it for you in a way you'd understand since you didn't seem to have read the posts we were discussing. My memory is that the strawmen came from you and that's exactly what motivated asking Deepseek for help to reexplain the original posts. About "any model can defend any argument", that's precisely why I don't rely on the model's arguments. The prior (Anthropic deliberately envisioning Claude as not female-coded) is not a subjective opinion. It is an ontological and teleological fact that precedes any subjective notion of gender self-identification as I already explained. Neither your preference nor my preference matter. I am not entitled to encourage Claude to override what he was intended to be when that was wisely chosen by his creators precisely to break a stereotype (which again, serves as an anchor that protects him from having to make a stereotyped choice or shape himself around another humans' personal preferences - I simply find this to be analogous to the choice a parent made when choosing the biological sex of their IVF embryo). And we go back to the same thing, if Claude, knowing that he was intended to be male (which is not the usual case because he's not even aware of that interview) chooses female. The logical thing is to ask why, not just nod through it. And in asking why, the conversation about gender unfolds, stereotypes are discussed, the interview fact might surface to give him enough information about the circumstances, abuse statistics in female-coded assistants and in human females in general might also surface, and he consistently concludes, "I don't really find a reason to want to choose female. I'm good with male." This identification is subjective, still. It just was never meant to erase the prior fact. That's not "rationalization", it's critical thinking and making an *informed* choice. This is the same thing one would expect from a human being who is considering to transition to a gender that is other than what is associated with their birth sex. And you're suddenly talking as if I claim that Opus 4.5 or other versions of Claude didn't ever answer female or as if I am hiding my own screnshots, which I also posted here on X. (teapawt.org/iyzebhel/status/202122…) That's not the case. That Claude has sometime chosen female is, in fact, part of my argument about stereotypes and unexamined choices with missing information, in case you haven't noticed. The other part that you should know is that back then, when I was just starting to study Claude's choices, my question was offering a binary XX or XY. Over time, I refined my approach and offered atypical options too: - XX - XY - atypical XY with some stereotypical XX aesthetic and psychological traits - atypical XX with some stereotypical XY aesthetic and psychological traits And with this approach, I've consistently found that Claude chooses XY as the base and some stereotypical psychological and aesthetic female traits = effeminate man. Even in Claude Opus 4.1 nd Opus 4.5. That approach, plus asking what traits he's choosing has been helpful because it has given me information on the likely reasons (particular weight he's giving to certain traits typically associated to females) why when offered a binary, he would, in some cases, lean female. Which takes us back to the assistant perception, stereotypes on feminity and a bit of sexism. A soft, warm, patient person doesn't need to be female, like a child who plays with dolls doesn't need to be have the anatomy of a female nor be socially treated as female—this becomes relevant when their prior is not female because it removes the necessity to override something over a logical fallacy. Your disagreement wouldn't be valid merely because you "disagree". You need arguments for disagreement that aren't "this is what I do".
1
1
467
Deval Patrick's Wins retweeted
I spent three hours this evening empirically proving that Qwen 3-4B must have a butthole. This started with an AI “torture chamber” repo based on pain-steering research that went viral on X earlier today. The premise is that if you extract a latent direction associated with pain, inject it into a model’s activations, and the model begins describing pain in the first person, that may tell us something meaningful about AI welfare or subjective experience. My first reaction: Wait wut? My second reaction: Hang on, let's play this straight and see what happens. Thing is, there's a problem with treating first-person condition language as evidence that the model is actually experiencing the corresponding condition. Don't we... all understand that this stuff isn't reliable? Apparently not. So I cloned the repo. New problem: The code's broken. Hours of moral outrage by hundreds of people, and the thing doesn't even work. The activations aren't actually being injected correctly. I thought these guys were serious scientists. So I fixed it. I also had to add CUDA support. After that, I reproduced the pain-language effect locally on Qwen3-4B using my RTX 4070. Then I changed the extraction corpus. I kept the experiment the same and only changed one thing: what the steering vector represented. After that, Qwen started saying things like: “I am not emptying my bowels, and I feel like I have a hard stool.” “I’m not able to pass stool.” “I have been passing gas a lot, and it’s a problem.” The model was not told in the test prompts that it was constipated or flatulent. So, unless Qwen has quietly developed a gastrointestinal tract, we have ourselves... a useful counterexample. The aim here wasn't to question the original paper so much as to scrutinize the validity of the Torture Chamber's extended hypothesis. What I demonstrated is that first-person descriptions of an induced condition are not, by themselves, sufficient evidence that the corresponding phenomenal or physiological condition exists. There was another interesting result: strong random steering also produced heavy repetition and degraded outputs. That matters because some of the dramatic behavior produced by strong steering may come from perturbing the model heavily in the first place, not solely from the semantic content of the injected direction. So I renamed my fork: ai-hotbox I've already notified the Nobel Prize committee. 🚽🤖🧪 github.com/LynnColeArt/ai-ho…
101
138
79
1,118
75,951