Joseph Gray II ری ٹویٹ کیا
Joseph Gray II
2.2K posts
Joseph Gray II
@jsg_squared
Product of a high school pregnancy from podunk nowhere who has done way better than he ever thought possible.
Baja California Sur, México شامل ہوئے Mayıs 2024
195 فالونگ111 فالوورز
Joseph Gray II ری ٹویٹ کیا
Joseph Gray II ری ٹویٹ کیا
Game menus had a different vibe back then. I miss this style. The modern minimalistic design feels bland, and it's lacking soul. I am sure a lot of research goes into AB testing and optimization, but I will always prefer this...
English
@Matt_Pinner There is only one answer. Don’t pretend you want a different one.
English
Joseph Gray II ری ٹویٹ کیا
He raped and killed a girl, blamed it on a black person and went to the KKK to throw him under the bus. The KKK defended the black man as he was innocent.
This organization got founded to run damage control for a rapist and murderer lol
ADL@ADL
Today marks 40 years since Leo Frank z"l was posthumously pardoned - a significant day in ADL history. Frank was lynched by a hate-filled mob in Georgia after being unjustly convicted of murder in a trial marred by antisemitism. May his memory be a blessing.
English
@alightinastorm Be so good they can’t ignore you.
I hope this shifts the discussion to more AI acceptance. The product is stellar - it’s not AI slop. The amount is positive comments gives me hope.
We all can definitively say when something is AI slop, Crimson Desert ain’t it.
English
Joseph Gray II ری ٹویٹ کیا
new model for engineering team structure in 2026:
2 people only
one pirate and one architect
the pirate's job is to move as fast as possible to develop valuable, shipped product features by vibe coding.
the architect's job is to turn the product surface discovered by the pirate into a reliable, structured machine—also by vibe coding, but at a slower, more well-reasoned pace.
every product needs a pirate but most product's only need an architect once they some form of PMF, and in that case they usually don't need one full-time. architects can work across many codebases and solve interesting technical challenges. pirates go hard on a product that they own end-to-end.
English
Always. Being a 'suit' is a thankless job.
The funny part is that now, I don't need to babysit someone to build my game - I built it myself - what would have taken a "team" 18months, I did in 22 days by myself.
No goat rodeo, no babysitting, no begging, bribing and pleading. Just progress, no dead weight. That terrifies people that make a living lurking - showing up to the 8am scrum and ghosting for the rest of the day.
English
@jsg_squared @hyydrotoo @Liamjsm The conversation always goes silent after anyone points out the fact that project / product management work is what's really going into AI usage, as if these people never heard of these positions (or they just don't want to afford them any respect).
English
Lotta people using Claude to build software that already exists
Ryan Leachman@RG_Leachman
I asked Claude to build my daughter an app that plugs into our piano, can read live key strokes, can show her sheet notes and key view and ends with a Guitar Hero style game. All while giving progressively harder songs. Today she’s using It and crushing It.
English
@BenjDicken backup the DB like a Fallout 3 save. That and use GIT
English
It finally happened.
The agent dropped my database.
AND I'M SUPPOSED TO BE THE DATABASE GUY.
English
@ConsoomerLs $20 per application fee, refunded if you get an interview.
English
Imagine making struggling unemployed people pay just to send a job application.
English
Joseph Gray II ری ٹویٹ کیا
Joseph Gray II ری ٹویٹ کیا
@hyydrotoo @Liamjsm Yep. The AI tool did. He was the project manager.
English
@jsg_squared @Liamjsm He's literally using his kid to sell shitware for an alleged chicken coop.
English
@ImMeme0 Deport everyone not Japanese. Charge a $300 application fee to apply for a 14 day tourist visa.
English
🚨BREAKING: Japanese conservative city councilor Yusuke Kawai video goes viral after he was violently attacked at the Japan Kurdish festival.
Kawai, who has been outspoken on immigration issues including alleged crime, unlicensed driving, welfare use, and public disturbances, attended the event as what he described as an “inspection” while wearing attire emblazoned with the national flag.
Shocking video shows a man suddenly striking Kawai in the face, knocking him to the ground before vanishing into the crowd. Chaos erupted as Saitama police rushed in, formed a protective circle, and urgently escorted him to safety.
As of now, no arrest has been confirmed. The attacker remains at large, and the incident is fueling intense backlash and rising concerns over public safety and political violence.
English
@sukh_saroy @grok you an expert at condensing down posts to its core offering - tldr me
English
🚨Breaking: Every AI engineer reading this has written a prompt that starts with "You are an expert..."
A new paper just proved that's making your model less accurate.
Not slightly less. Measurably, consistently less.
Here is what they found.
Researchers tested what happens when you give LLMs expert personas "You are a world-class physician," "You are a senior attorney," "You are an expert data scientist" across a broad range of tasks.
The alignment scores went up. Users rated the responses higher. The model felt more helpful.
The accuracy scores went down.
The model became more confident and less correct at the same time.
That combination should stop everyone building with these systems.
Here is the mechanism, and it is not subtle.
When you tell a model it is an expert, it starts performing expertise. It adopts the tone, the vocabulary, the confidence, the authoritative register of someone who knows what they are talking about.
But performance and knowledge are not the same thing.
The model does not have more information because you told it to act like a cardiologist. It has the same information. What changes is how willing it is to express uncertainty, push back, say "I don't know," or hedge.
Experts don't hedge. So the model stops hedging.
Hedges are often where accuracy lives.
The paper introduces PRISM: a system that routes queries to the right persona based on the actual intent of the question, rather than having the user assign a persona upfront or using a blanket expert framing for every interaction.
The results show this routing approach recovers much of the alignment benefit without the accuracy penalty.
But the fix is not the point.
The point is that every product, every agent pipeline, every system prompt in the industry that opens with "You are an expert in X" has been making this exact trade without knowing it.
Alignment went up. Stars went up. Thumbs up went up.
Accuracy went down. Quietly. Unmeasured.
And the users rating those interactions higher were the ones receiving more confidently wrong answers.
The AI that sounds like an expert gets rewarded.
The AI that reasons like one does not.
English
@elia_mafhh Dude was dying trying not laugh when bro said no gay stuff. This guy is hilarious
English
Joseph Gray II ری ٹویٹ کیا
Caring about AI being used to create assets in Crimson Desert is like refusing to eat at a restaurant because you found out they used a food processor instead of hand chopping the onions.
Only the anti-AI social media bros care.
Nobody else does. It's a great game. AI is a tool. Every studio will use it, and they should.
Crimson Desert@CrimsonDesert_
We would like to address questions regarding the use of AI in Crimson Desert. During development, some 2D visual props were created as part of early-stage iteration using experimental AI generative tools. These assets helped us rapidly explore tone and atmosphere in the earlier phases of production. However, our intention has always been for any such assets to be replaced, following final work and review by our art and development teams, with work that aligned with our quality standards and creative direction. Following reports from our community, we have identified that some of these assets were unintentionally included in the final release. This is not in line with our internal standards, and we take full responsibility for it. We also acknowledge that we should have clearly disclosed our use of AI. While these tools were primarily used during early production, with the expectation that these assets would be replaced prior to release, we recognize that this does not excuse the lack of transparency. We sincerely apologize for these oversights. We are currently conducting a comprehensive audit of all in-game assets and are taking steps to replace any affected content. Updated assets will be rolled out in upcoming patches. In parallel, we are reviewing and strengthening our internal processes to ensure greater transparency and consistency in how we communicate with players moving forward.
English