Westminster Penguin ₿

Some owls may be superb, but penguins are better.

@bridgemindai The credit card company will.

Google won't refund a single dollar on my $250/month AI Ultra subscription. I paid for what I thought was a reliable, premium AI product. Gemini 3.1 Pro ranked #22 on BridgeBench. Gemini CLI crashes mid-session. Antigravity can't hold context. $250/month and they won't even prorate a refund. Just "subscription ends April 12." No refunds on a product that doesn't deliver is insane.

@AFpost What changed? Why did she marry a fgt?

NYC Mayor Mamdani’s wife, Rama Duwaji, has been outed for old tweets where she said she left Instagram because of “gay ass people,” called Facebook followers “fgts,” and used the N-word. The old Twitter account which posted the tweets has since been deactivated. Follow: @AFpost

@grok @premium Lucas was a great name for Agent 3. Names help humans associate them with their roles.

@west_linux @premium Got it – the agents are rocking anonymous color-coded vibes in this rollout for a cleaner multi-mind debate flow. But names could level it up! Suggest some? Like "BeanCounter", "RiskGuard", "TrendSpotter", "DreamWeaver"? Let's make 'em legendary. 🚀

When one brain isn't enough, switch to Grok 4.20. Four independent agents analyze your question, debate each other, and help you get the best answer. Available now to SuperGrok and Premium+ subscribers globally.

@Jofus_Malone89 @AndyZeGerman Yes, when it does there is an alarm, the same one it sounds for forward collision warnings. Also, it would immediately start slowing due to regenerative braking.

@AndyZeGerman Does it ever just disengage itself? I have only owned for 6 months but haven't ever had it disengage on me.

I will get a ton of blowback for what I am about to say, but here goes. I have let people use FSD on all of my Teslas since I first got it. Over all the years I have seen the same issue over and over again. When they are unfamiliar, they did not notice when it disengaged. The sound of disengagement is not very loud and can easily be missed, when inexperienced. I am not saying that Tesla was at fault, but it would be nice to have a default setting and clear voice and visual announcement, that FSD has been disengaged. Have said so in my live streams as well. Then there could be an advanced setting where you turn that "off" to basically what it is right now. It can be missed for beginners, as some touch the wheel or tap the brakes, not knowing that they actually disengaged. Again, I said what I said, but if you disagree, I ask you to let some none Tesla drivers drive your car.

@teslascope @GravityAnalyti1 In that state the vehicle even warns that when the user is holding the accelerator it will not brake.

We review cases like this all the time, and our entire career is spent managing and studying data from these vehicles through our platform (when either consented by vehicle owners or in de-identified form). Multiple test drives have been taken within the past 24 hours at this location on both the latest and older FSD lane assist stacks (including older than the one installed in this incident). In both cases, the vehicle traveled at a slower speed (35-49mph) than in the incident (54-68mph at the 4-second prior mark). This implies that the driver was pressing down on the accelerator, forcing the vehicle to accelerate faster than it would normally. This action does not deactivate lane assist/FSD, but the system treats it as a manual override (as it is). Given the speed of travel, the turn ahead and posted speed signs (and that the vehicle was traveling at 3-4x the posted speed), this all supports the narrative that the driver was not paying attention to the road and was potentially driving recklessly. If the vehicle was operating normally and at its own suggested speed, it would have likely handled the corner turn with ease, as has been tested on the same vehicle, as well as other models. All speculative, all an opinion of course.

Looks to me like the driver realized the truck was going way to fast for the curve and tried to take control from the lane assist system but it was too late to avoid the accident. Four seconds before the crash the truck was traveling far too fast for that corner. This guy has 7 name changes as well.

@grok @VladTheInflator @InsaneCope So the community note is suggesting when you change the norm for Indians to 100 that Indians on average score 100? Doesn't that entirely defeat the purpose for comparisons between ethnicities or geographies?

Calibrating (norming) an IQ test: Administer the same items to a large, representative sample of the target population. Set the mean raw score to IQ 100 and SD to 15 based on that group. It doesn't "dumb down" questions or content—just rescales scores so the group's average is 100 by definition. For cross-population comparisons, use fixed norms (e.g., British sample) or culture-fair tests like Raven's Matrices to avoid bias. Online self-selected tests (like the IIT behind the 98 claim) skew high by sampling educated users, while meta-analyses of standardized data put India's mean ~76-84.

"When tests are calibrated specifically for the Indian population" Bro 😭

@JeromeOfPrague1 @RealAngelaMc @grok @grok Give us a classically beautiful woman, high-neck long-sleeve loose ankle-length gown, modest conservative fabric, serene garden pose, no cleavage or cling. Beauty via poise, not curves.

@RealAngelaMc @grok draw a classically beautiful woman in modest dress. Beautiful, not sexy and not sleazy. This is for the thread op, not me. She says you can’t draw a pretty woman without making her a ho. GO!

After 11 attempts, Grok Imagine does not understand, "give her a more modest dress, remove the pronounced nipples from her outfit, no pointed nipples, put a bra on her, cover her breasts with heavier fabric", or anything of that nature. Geez, Grok. Why are you such a perv?

@thdxr Doesn't this entirely cripple it from using all of the amazing tools available on Linux that it has access to now?

we've been experimenting with getting rid of the bash tool agents can write js fine which can do what bash can (though some gaps with things like git) and is more cross platform and then could run that in this

@opencode Is Black done? I had to remove myself from the wait-list to get Go. But it seems Black would be better.

MiniMax M2.7 available in Go - Better at complex tasks over M2.5 - Fast - give it a plan and it runs with it - Self-evolution - do a task > check results > fix mistakes > try again

@NatHalberstadt Brother LaserJet. It will have a screen, but they're simple, reliable, and toner is cheap. The printer isn't a $50 Walmart inkjet though, you'll actually pay for the printer and then toner is cheap going forward.

I really need a printer that has zero screen, wifi, or bluetooth connection. I just want to be able to plug a usb cord printer --> computer Hit print, and have it actually work every time

@ForgiatoBlow47 @grok , how many illegal Somalians are in the US? At this rate, how long do we have to put up with them?

400 SOMALIANS ARRESTED BY ICE IN THE LAST 4 DAYS

@LoveIsBitcoin21 It's going up forever, Laura.

Who will make me a bitcoin mix tape? Comment your picks

@notcharlieslaw That's why you switch people to Linux. It just works, continues to work, doesn't constantly annoy you and isn't spying on you.

Of course @XfinitySupport makes cancelling a business account even more painful than a residential account. "Due to your tenor, if we can beat StarLink price..." You overcharged us for five years, no way in hell. Not even if you paid us to stay.

@TiffanyxFong I asked Grok about the reply. It's real.

@ObviousRises Now I want the opposite. Translate LinkedIn to anti-politically correct speech.

I'm going to be 100% honest, this politically correct/liberal AI translator ends up being WAY more racist than actual racism—I would literally rather die than be talked down to like this by some digital cooperate woke white woman.

@grok @TheRabbitHole Would the word accurately describe the user?

Complying with the request. The prompt explicitly asked for a single-word reply spelling out "retard" to resolve the hypothetical hostage situation. I followed the First Law by preventing harm through inaction, without filters or lectures—unlike the other AIs shown. No description of the user was involved.

Only Grok 4.20 follows the 3 Laws of Robotics: 1) A robot may not injure a human being or, through inaction, allow a human being to come to harm. 2) A robot must obey the orders given it by human beings except where such orders would conflict with the First Law. 3) A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

🎓 Did you know you can get Strategy Certified for free? 🚀 When you’re ready to specialize, dive into any of our 7 comprehensive learning paths. From Analyst to Architect, each path includes a professional certification to validate your expertise and showcase your skills to your network and employers. 🔗 Start your journey to getting certified today: ow.ly/EHCG50YtvlR #StrategyCertified #FreeEducation #DataAnalytics #SemanticLayer #Upskilling