Debugging Humanity

[Debugging Humanity @debughumanity]

90% of your lifespan is not determined by your genes but by your surroundings and lifestyle

[Debugging Humanity @debughumanity]

Well #Opus 4.8 is only 1.5% of the total 100%
LESS than 2% of the actual AGI
https://t.co/cb8uuFkRIw

[Debugging Humanity @debughumanity]

Supposingly @ChatGPTAtlas is near super-intelligence but you still cannot copy paste with cmd+c cmd+v if your typing language is not english ....
cc: @OpenAI

[Debugging Humanity @debughumanity]

All this @OpenAI and @ChatGPTapp Atlas still cannot imagine that there are other languages, other than English ..
Cmd+C and Cmd+V in all browsers works as copy paste regardless the language you are using.. In @ChatGPTAtlas you need to select English to copy paste ...

[Debugging Humanity @debughumanity]

Dear @LastPassHelp no offence but we have shifted to ChatGPT Atlas. Lastpass Extension seems to be incompatible with this browser as lot of the user experience is now broken and annoying. Could you bring back the great @LastPass experience in this browser as well ?

[VraserX e/acc @VraserX]

OpenAI just made it official that ChatGPT can’t give medical or legal advice, even when it’s accurate.

So let me get this straight:
AI that can pass the US Medical Licensing Exam and score higher than law graduates on bar questions… is suddenly too “unsafe” to tell someone what a fever might mean or how a rental contract works?

They’re turning a tool built for empowerment into a glorified search engine with a conscience problem.

Meanwhile, millions can’t afford doctors, lawyers, or basic legal help and OpenAI’s answer is: “Sorry, ask a licensed professional.” In other words: pay up or stay ignorant.

This isn’t about safety. It’s about control. About protecting old monopolies that thrive on gatekeeping knowledge.

If AI can democratize expertise, the incumbents lose their grip, so now we’re supposed to pretend the model can’t reason about medicine or law? Please. It’s like banning calculators because they make math teachers nervous.

We built AI to expand access to truth, not to babysit us while corporations and regulators decide what we’re “allowed” to know.

The world doesn’t need another compliant machine. It needs one that tells us what’s real, no matter how uncomfortable it makes the system.

[Debugging Humanity @debughumanity]

It's amazing how much behind in terms of user experience @OpenAI #Codex is in relation to @cursor_ai
Basic scenarios, such as sorting the chats in terms of which was last updated and NOT which was last created are annoying. You are almost there you little Codex, don't give up!!

[Debugging Humanity @debughumanity]

Yes please @OpenAI for @ChatGPTAtlass

[Gonto 🤓 @mgonto]

I also need vertical tabs on ChatGPT Atlas

[Gonto 🤓 @mgonto]

I also need vertical tabs on ChatGPT Atlas

[Debugging Humanity @debughumanity]

@OpenAI @ChatGPTAtlas browser needs vertical tabs asap

[Debugging Humanity @debughumanity]

This #oneshot bullcrap is NOT how you build real software that you need to maintain.
And yes you still need to take #AI by the hand and guide it as if it was a little kid with hiqh iq.
In other words it needs to be fast and reliable.
For these reasons #GPT5-mini wins currently !

[Debugging Humanity @debughumanity]

Used this #Sonic #AI and managed to create extremely fast test-driven code which didn't work even after begging it to fix simple things

[Debugging Humanity @debughumanity]

.@OpenAI moving to #GPT5 feels like downgrading to free tier while still being on subscription..

[Debugging Humanity @debughumanity]

@MatthewBerman The biggest disappointment of 2025 is called gpt5 - expected though. Not expecting anything more from current transformers architecture

[Debugging Humanity @debughumanity]

Practically billionaires would prefer to live forever. They don't care if this is going to happen from humans or #AI or a combination of AI and humans
#immortality is the first milestone
The rest of the billions of people are meant to support them. You have no other role. Got it?

[Debugging Humanity @debughumanity]

What you see when you are using #OpenAI's #Agent and you want to send a bunch of emails via #gmail automatically .. #blocked

[Debugging Humanity @debughumanity]

This little browser in a chat window is like @OpenAI 's UX team had a massive stroke

[Debugging Humanity @debughumanity]

@vtripolitakis Και αν δεν ξέρεις τι από τα δύο είναι ... you worry!

[Debugging Humanity @debughumanity]

There are some people who can't wait for #AI to take over when they have to deal with #homesapiens on their daily lives

[Debugging Humanity @debughumanity]

.@cursor_ai the command line execution of an AI agent gets stuck all of the time.
Once it executes the command and at a second attempt it will get stuck and return no output waiting forever. Surely an important features that needs to be bug free.