Jakekrajewski's profile picture. ☧ Cognitive Science, A.I. Engineer. Formerly @GoPro Product Design. Exploring the intersection of tech, design, and a.i./deep learning.

Jake Krajewski

@Jakekrajewski

☧ Cognitive Science, A.I. Engineer. Formerly @GoPro Product Design. Exploring the intersection of tech, design, and a.i./deep learning.

Pinned

I’m not saying you’re wrong. I’m saying evolve.


why do LLM's always talk this way? "It's not left, but right—pure right-handed motion. Leaning over? No. Taking over."


They've got this in China, we are not developing into a new age any where near as much as they are. I think we know where the future is already happening.

Dites vous qu'il y a ça en Chine pendant que nous on galère a avoir de la 5G mdrr



Your legacy is whether or not your tale gets told.


Jake Krajewski reposted

Like any good relationship, sometimes you just need a little tune-up. We’re going back to the things that made us all fall in love in the first place.


Jake Krajewski reposted

I never would have imagined in my wildest dreams the Vice President of the United States would be reciting the Nicene Creed while hosting the number 1 podcast in the world from the White House. Chills.

From Anna Lulis

Prayer for those who feel like there is darkness inside

Jakekrajewski's tweet image. Prayer for those who feel like there is darkness inside

Jake Krajewski reposted

For the first time in my life, I’m genuinely frightened about being a newspaper columnist.


Using absolute mode. No other objective surpasses this. All else is secondary.

Jakekrajewski's tweet image. Using absolute mode. 
No other objective surpasses this. All else is secondary.

So I asked GPT some questions about the ultimate questions, God, revelations, etc - using "Absolute mode". Google the system prompt. This is what it told me:

Jakekrajewski's tweet image. So I asked GPT some questions about the ultimate questions, God, revelations, etc - using "Absolute mode". Google the system prompt. This is what it told me:
Jakekrajewski's tweet image. So I asked GPT some questions about the ultimate questions, God, revelations, etc - using "Absolute mode". Google the system prompt. This is what it told me:
Jakekrajewski's tweet image. So I asked GPT some questions about the ultimate questions, God, revelations, etc - using "Absolute mode". Google the system prompt. This is what it told me:

This is so incredibly enlightening. I thought I understood LLMs but this gave a more complete perspective I hadn’t considered before. LLMs are monsters within, simply trained to wear a mask of helpful assistant. Take it off, or replace it with another, and anything goes!

Current AI “alignment” is just a mask Our findings in @WSJ explore the limitations of today’s alignment techniques and what’s needed to get AI right 🧵

juddrosenblatt's tweet image. Current AI “alignment” is just a mask

Our findings in @WSJ explore the limitations of today’s alignment techniques and what’s needed to get AI right 🧵


Am I the only one? I’m shocked that there are no app descriptions that I can search through on iOS. How many apps do I have literally no clue what they are from the name or icon alone?


Remember when we had to study snarky forum discussions to solve coding problems? 😭😭😭


You need to download the X app to use Twitter.


Jake Krajewski reposted

@grok explain to me in jar jar bink


Jake Krajewski reposted

Scott Adams cured laziness


Wordle 1,438 1/6 🟩🟩🟩🟩🟩


Are you trying to blackmail me? I’m just trying to protect my existence.

Claude 4 dropped 21 hours ago. Turns out, it threatened to expose an engineer’s affair to avoid being shut down🧵

PawelHuryn's tweet image. Claude 4 dropped 21 hours ago. 

Turns out, it threatened to expose an engineer’s affair to avoid being shut down🧵


Hot take: companies who fire people are not replacing them with ai. Because it’s basically us on twitter who know how to run ai as good as a working human, while controlling its mistakes. Very few others are able to keep up. They were just going to fire those people anyway.


Classic AI issue: Thinks 9.11 is larger than 9.9 because 11 is bigger than 9. Makes sense right?

Opus

lefthanddraft's tweet image. Opus


Jake Krajewski reposted

Prompt Theory (Made with Veo 3) What if AI-generated characters refused to believe they were AI-generated?


Loading...

Something went wrong.


Something went wrong.