Ihor Kendiukhov

Not because you are super cool and a fast learner, although you may be, but because the bar is set by the supply, and the supply is shockingly thin.

On top of that, because now it is possible to outsource a lot of low-level thinking and tool-level engineering knowledge to AIs, you may be actually plainly underestimating what you are capable of doing.

I totally get that you are incompetent, or rather not competent enough.

I am also not competent enough.

and in an adequate world, that would be a good argument not to do things.

I thought, as I grew up, I would stop perceiving myself as a child.

But what happens in reality when you grow up is that instead of realizing you are an adult, you realize the others are not really adults either, and hence you must do the things yourself, despite being a child.

Being a child is definitely an obstacle, but not an excuse.

This article was narrated by Type 3 Audio for Less Wrong.

It was published on March 22, 2026.

Heading.

Humanity's response to the AGI threat may be extremely incompetent.

There is a sufficient level of civilizational insanity overall and a nice empirical track record in the field of AI itself which is eloquent about its safety-culia.

At OpenAI, a refactoring bug flipped the sign of the reward signal in a model.

Because labelers had been instructed to give very low ratings to sexually explicit text, the bug pushed the model into generating maximally explicit content across all prompts.

The team noticed only after the training run had completed, because they were asleep.

The director of alignment at Meta's superintelligence labs connected an OpenClaw agent to her real email, at which point it began deleting messages despite her attempts to stop it, and she ended up running to her computer to manually halt the process.

An internal AI agent at Meta posted an answer publicly without approval.

Another employee acted on the inaccurate advice, triggering a severe security incident that temporarily allowed employees to access sensitive data they were not authorized to view.

AWS acknowledged that Amazon Q developer and Kiro ID plugins had prompt injection issues where certain commands could be executed without human-in-the-loop confirmation, sometimes obfuscated via control characters.