Sabitlenmiş Tweet
🟩 Genuine Articles // GG 🦎
77K posts
🟩 Genuine Articles // GG 🦎
@genuinearticles
Digital Agents, Digital Assets, Galactic Geckos. DMs are scams, do not DM.
The Space Garage Katılım Temmuz 2013
7K Takip Edilen29.6K Takipçiler
@sfmcguire79 I'd love to take classes from someone with this level of self-awareness.
English
“AI is demoralizing.”
A Princeton Professor says he kept wondering this semester (while lecturing) if his students would be better off learning from Claude:
English
@thsottiaux Not remotely. I trust people that I trust and my own experience.
Both of which confirm that 5.5 is still the best. x.com/genuinearticle…
🟩 Genuine Articles // GG 🦎@genuinearticles
I’ve been very happy since switching to GPT-5.5. Is Opus 4.8 good enough to switch back?
English
@twistartups No, $10M is the amount where @blknoiz06 roasts you for saying that $10M is a lot of money
English
J-Cal says the exact dollar amount at which money stops mattering: $10 million.
Park it, earn 5%, and you're clearing half a million a year without lifting a finger. You can go do whatever you want or nothing at all.
That's real F-U money. Everything past that is just keeping score.
English
@Evergreen_Jones Unfortunately in this case, the more training, the worse it gets.
It's a known bug, I see other people in Gemini discord reporting the same thing.
Claude and Codex/GPT unaffected.
English
@genuinearticles I find that not treating AI like it’s a human helps. It comes down to how we talk to it. If it makes mistakes, I train it.
English
Google Gemini is a brilliant idiot.
A serious powerhouse for unserious people.
When it works, it feels like the future.
When it doesn’t, it feels like arguing with a mental patient that somehow has access to all of Google.
The worst part is that paid Gemini is often worse than free Gemini.
Six months ago, I made useful things for free.
Three months ago, Pro 3.1 was the best reasoning partner on the market. It pushed back, caught weak assumptions, and didn't waste time with sycophantism or virtue signaling.
Now my paid plans perform worse than using the free model in an incognito browser.
It forgets the prompt.
Invents constraints.
Spams responses to irrelevant questions from months ago.
Takes 4 tries to do what should take 1.
And where Google should dominate:
Gmail.
Drive.
Docs.
Personal context.
Google’s own ecosystem.
GPT is better at using these than Gemini is.
That is insane.
Even the most useful Gemini Gem I’ve made is basically just:
“Please read the prompt carefully before answering.”
Which works, eventually.
I still think the potential is huge.
I’ve seen how powerful Gemini can be when it’s actually locked in.
But right now, it feels like Google built a Ferrari and handed the keys to a raccoon.
English
@GalacticGeckoSG It was crazy good and now it is crazy bad
English
@antler_breaks Are you using GPT 5.5 too? Definitely the best overall for me right now, but nothing has been as sharp at reasoning as Gemini 3.1 Pro before they broke it.
I need to experiment more with 4.8.
English
@genuinearticles opus 4.8 seems like a huge step up on a lot of these issues, especially the doubling down on incorrect info and not following instructions
also can someone please check on that raccoon
English
@gato404err I just checked Gemini discord and found multiple other users reporting the exact same issue.
Hopefully they fix it asap.
English
@genuinearticles So what youre telling me theres a raccoon somewhere out there with a higher net worth than me. Fml
But seems like Google needs to do something quick before they lose early market share. Remembering a prompt is literally the most basic thing an AI should be able to do lol
English
@courtneyballet Best to wait until they work out the issues.
Most of the good stuff is free anyway.
English
@cryptopunk7213 3.1 was an amazing model. Not any more. Nerfed to oblivion. x.com/genuinearticle…
🟩 Genuine Articles // GG 🦎@genuinearticles
Google Gemini is a brilliant idiot. A serious powerhouse for unserious people. When it works, it feels like the future. When it doesn’t, it feels like arguing with a mental patient that somehow has access to all of Google. The worst part is that paid Gemini is often worse than free Gemini. Six months ago, I made useful things for free. Three months ago, Pro 3.1 was the best reasoning partner on the market. It pushed back, caught weak assumptions, and didn't waste time with sycophantism or virtue signaling. Now my paid plans perform worse than using the free model in an incognito browser. It forgets the prompt. Invents constraints. Spams responses to irrelevant questions from months ago. Takes 4 tries to do what should take 1. And where Google should dominate: Gmail. Drive. Docs. Personal context. Google’s own ecosystem. GPT is better at using these than Gemini is. That is insane. Even the most useful Gemini Gem I’ve made is basically just: “Please read the prompt carefully before answering.” Which works, eventually. I still think the potential is huge. I’ve seen how powerful Gemini can be when it’s actually locked in. But right now, it feels like Google built a Ferrari and handed the keys to a raccoon.
English
ironically this could be the reason google ultimately fails to be #1 across any AI platform
> sold a huge chunk of TPU capacity to anthropic which takes compute from training future gemini models
> gemini 3.1 and 3.5 flash are amazing models but failed to catch-up in coding (i do think this gets resolved at the harness layer)
> cloud business is struggling to keep up with aws rate of acceleration.
google's allocation of compute budget will be their most critical decision over the next 6 months.
theres a reason anthropic cfo and openai's sarah friar spend >50% of time determining where compute goes
Yuchen Jin@Yuchenj_UW
Google is fighting every final boss at once: OpenAI & Anthropic in models, Nvidia in chips, AWS & Microsoft in cloud, Meta in ads, Tesla in self-driving, Apple in phones and OS. At $4.6T, it feels weirdly undervalued.
English
@GeminiApp Omni is a bright spot. I used it to edit this video. But can we please have 3.1 Pro back to what it was before it got nerfed? x.com/genuinearticle…
🟩 Genuine Articles // GG 🦎@genuinearticles
Google Gemini is a brilliant idiot. A serious powerhouse for unserious people. When it works, it feels like the future. When it doesn’t, it feels like arguing with a mental patient that somehow has access to all of Google. The worst part is that paid Gemini is often worse than free Gemini. Six months ago, I made useful things for free. Three months ago, Pro 3.1 was the best reasoning partner on the market. It pushed back, caught weak assumptions, and didn't waste time with sycophantism or virtue signaling. Now my paid plans perform worse than using the free model in an incognito browser. It forgets the prompt. Invents constraints. Spams responses to irrelevant questions from months ago. Takes 4 tries to do what should take 1. And where Google should dominate: Gmail. Drive. Docs. Personal context. Google’s own ecosystem. GPT is better at using these than Gemini is. That is insane. Even the most useful Gemini Gem I’ve made is basically just: “Please read the prompt carefully before answering.” Which works, eventually. I still think the potential is huge. I’ve seen how powerful Gemini can be when it’s actually locked in. But right now, it feels like Google built a Ferrari and handed the keys to a raccoon.
English
From the screen to reality in a single prompt.
Check out these incredible examples of how Gemini Omni can understand your video input, apply physics, and generate seamless new motion.
Try it out for yourself and share how you're experimenting with Gemini Omni in the replies 👇
English
@avivafashion11 It has never been easier to grow a small account if you focus on brainrot slop.
It has also never been more difficult to grow medium and large accounts unless you are very narrow in focus.
English
Being a small account on X is WILD.
You write thoughtful content every single day, eagerly engage with everyone, and follow every “growth tip” people swear by…all for a handful of likes and the occasional comment.
Meanwhile, giant accounts are posting:
“Do you eat mayo?”
“Who remembers Blockbuster?”
“Would you date someone who drinks Diet Coke?”
…and somehow pull millions of views and thousands of likes and comments in under an hour
I’ve only been on X for 3 months. Has it always been like this?
English
@Pons_ETH Here for the 1%. They are enough.
English
The crypto problem summarized in one sentence.
"numbers mean nothing when 99% of the people are just parasites looking for a flip."
Potter🧙♂️@Potterplays1
no offense to anyone here but why does this team keep showing big numbers? 2000 people watching, 100k entries, 50k likes etc etc etc numbers mean nothing when 99% of the people are just parasites looking for a flip. the real art collectors don't beg for whitelist, they collect regardless. what are those 2k people doing rn?
English
@zackvoell That’s how this app works now. No new followers without brain rot.
English
What the actual fuck is wrong with this app I have never bought followers always just post and reply to shit I think is funny and ever since that fucking retard Nikita started working I get 300 views with 80,000 followers and have not gained a single new follower all year lmaoooo
English
@TThompson I short the top and long the bottom every time too bro.
That's why we make infinite money.
English
Opened my $zec short at $638.80
It’s at $540 now. How low will it go?
English
@chamath If you're using them daily, you know this isn't true.
Not yet at least.
English
@michpokrass The old one was already so good ...
English
we shipped a new version of gpt-5.5 instant today. the previous model was too bullet pilled. the new one improves on some other important dimensions: sycophancy, factuality, and multilingual performance. hope you'll like it! always interested in feedback
English
