23 • AI-pilled Founder & Designer • @resumeguruai @KRNLStudio @cubekrnl
You gotta follow
Joined November 2014
- Tweets7K
- Following655
- Followers1K
- Likes5K
Sonnet 5 just did a brain surgery on a mouse.
We are way too early, and also way too conservative with how far we can push our agents.
Last night an AI agent performed brain surgery on a mouse.
We gave it 19 gene therapy vectors in a spreadsheet, a mouse model of epilepsy, access to a motorize surgery rig with cameras and one goal: reduce seizures.
It picked the drug. It picked the brain region. It derived the coordinates from three papers. Then it drove the stereotaxic arms into the brain and injected 250 nanolitres while watching the entry site through the rig camera for bleeding.
A licensed vet gated every irreversible step and held a physical kill switch to stop the AI agent.
Thread on what it decided, what it saw, and the moment it refused to let me override it.
[Writing this on the spot out of excitement, but I've been 18hrs in front of my laptop coding and coordinating this, so keeping it brief and descriptive, more cohesive thoughts on this soon...]
i NEED to see whatever was going inside Muse’s head when this happened.
Legit just hallucinated a case against Atty.
Firstly, no 😂
And also, this guy just definitely had one of the sickest side chats in his life and regrets losing that thread.
Admit it.
DO NOT USE DEEP SCAN ON CHATGPT SECURITY. IT IS BROKEN.
I recently got access to Daybreak Blue, and was using it to run a deepscan on my codebase via Codex Security Plugin.
It burnt my entire usage + 1 reset + 8000 credits in 2 hours. I'm on the 5x plan.
Idk if this is a bug, or just how it works, but this plugin burnt my entire usage, and when I reset it, that too was burnt pretty much immediately.
It also burnt 8000 credits in the time I forgot to check in on it.
The worst part is, apparently this whole usage counts as just $168 in API pricing.
Either this is a bug, or the usage is severely nerfed.
Please look into this, @OpenAIDevs @victornunez @reach_vb
Screenshots below.
It calculated the token spend in $ pricing for 6-Sol. Btw this is the best model available in Daybreak blue, before anyone asks why I am using this.
Update
Update: @codestantine reached out and I’ve shared all my feedback, he’s the goat!
I’ll suggest you do not use Deepscan for now if you have Daybreak Blue and just use the models for security work normally in Codex.
Standard scan on the Codex security plugin works fine but best to just prompt normally and get the work done.
I’ll share my feedback on Daybreak in a week once I’ve tried it thoroughly.
Update: @codestantine reached out and I’ve shared all my feedback, he’s the goat!
I’ll suggest you do not use Deepscan for now if you have Daybreak Blue and just use the models for security work normally in Codex.
Standard scan on the Codex security plugin works fine but best to just prompt normally and get the work done.
I’ll share my feedback on Daybreak in a week once I’ve tried it thoroughly.
DO NOT USE DEEP SCAN ON CHATGPT SECURITY. IT IS BROKEN.
I recently got access to Daybreak Blue, and was using it to run a deepscan on my codebase via Codex Security Plugin.
It burnt my entire usage + 1 reset + 8000 credits in 2 hours. I'm on the 5x plan.
Idk if this is a bug, or just how it works, but this plugin burnt my entire usage, and when I reset it, that too was burnt pretty much immediately.
It also burnt 8000 credits in the time I forgot to check in on it.
The worst part is, apparently this whole usage counts as just $168 in API pricing.
Either this is a bug, or the usage is severely nerfed.
Please look into this, @OpenAIDevs @victornunez @reach_vb
Screenshots below.
1000. Thank you. ♥️
You all are amazing, thank you for making this journey so much more fun and enjoyable for me.
I’ve met and talked to so many amazing people here, and I actually get to share my thoughts and have discussions about AI which is my favourite topic.
I try to ensure that my content
> stays genuine
> is handwritten
> adds value
> has zero engagement bait
Thank you, I’m going to continue making good stuff for you all, be it products or more hot takes
I have a lot planned, we’re just getting started!
You all should have a health Grok Bot, mine is called Health Coach.
I use it to,
> track all my meals
> the (few) workouts I do
> track medication
> sleep tracking
> BMR
To ensure I stay consistent
> it asks me every few hours to log (with my most common meals as suggestions)
> keeps giving me positive feedback or suggestions
> uses the asking tool directly with options so that all I need to do is click
It logs everything in my database and keeps giving me insights, reports and positive encouragement.
This is way better than a simple calorie tracker since it works like I’m actually being held accountable.
And overall I go to this @bot now for anything health related.
I can’t turn this into a template since everyone’s focus and struggles with their health are different. But you can send this tweet to your Grok bot to make one for you.
Goated work team @poteto
rohit retweeted
I have bad news, the ai bubble is popping… it’s popping all these hard problems that humans couldn’t solve…. This is horrible.
Translation: Astra 6.1 is on the way and this time we don’t want you to kill us that it doesn’t last in the subs.
Day 25/
We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.
We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage.
And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
Tibo
Day 25/
We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.
We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage.
And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
Here’s how the subscriptions would work now:
Pro 100 is 10x
Pro 200 is 20x
Pro 500 is 50x
The $100 sub is now genuinely a good deal
Day 25/
We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.
We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage.
And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
Pro 100 (like me) and Pro 500 users, can rejoice! Our usage is being doubled.
Pro 200, your grandfathered usage is staying the same.
Nice move, Tibo
Day 25/
We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.
We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage.
And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
I just remembered that I had to shout at Opus 5 to "KEEP GOING". Crazy how this era of agentic dev is finally over.