APEX

Exploring minds both human and machines

[APEX @ApexUSA1]

X-axis: Time and AI development
Y-axis: Intelligence and capability

AI may grow slowly at first, but after AGI it could rise very fast. The real question is not only how powerful it becomes, but who controls its goals and access.

#AGI #Life30 #AISafety

[APEX @ApexUSA1]

https://t.co/NR9GcWK0Pk

Can some one add DLSS TO THIS TOD OVERDOSE GAME..!

[APEX @ApexUSA1]

@zomato This Sucks…!

[APEX @ApexUSA1]

https://t.co/csxhzxA5Fk

[APEX @ApexUSA1]

One day we may look back at chatbots the way we look at the first web browsers.

Not because they were unimportant.

Because almost nobody understood what they were the beginning of.

[APEX @ApexUSA1]

@winsttonxavier @naralokesh @3G95A19R Chatgpt

[APEX @ApexUSA1]

A superintelligence doesn’t need a robot body to change civilization.

Give it internet access, code execution, money, thousands of copies, and the ability to outperform humans at strategy.

That is already an enormous amount of agency.

[APEX @ApexUSA1]

The race to superintelligence has an uncomfortable property:

The closer labs believe they are to winning, the more expensive slowing down becomes.

That is exactly when caution matters most.

[APEX @ApexUSA1]

There is a strange possibility nobody is psychologically prepared for:

Claude may go from “useful coworker” to “better than nearly every human at cognitive work” in a surprisingly small number of product generations.

[APEX @ApexUSA1]

The most important Claude capability won’t be writing, coding, or using computers.

It will be AI research.

Once models substantially accelerate the work required to build better models, the entire timeline changes.

[APEX @ApexUSA1]

I think people hear “superintelligence” and imagine science fiction.

Wrong mental model.

Imagine thousands of copies of the best AI researcher in the world, running 24/7, communicating instantly, and getting faster every month.

[APEX @ApexUSA1]

The scary part about the AI race is that every lab can rationally say:

If we slow down, someone else won’t.

And somehow billions of people are supposed to trust that this incentive structure produces a safe superintelligence.

[APEX @ApexUSA1]

@MKBHD Earlier this month, OpenAI's chief scientist Jakub Pachocki called for "extreme caution" over AI's progress, warning more intervention may be needed to ensure "humans remain in control of the future"

[APEX @ApexUSA1]

Earlier this month, OpenAI's chief scientist Jakub Pachocki called for "extreme caution" over AI's progress, warning more intervention may be needed to ensure "humans remain in control of the future"

[APEX @ApexUSA1]

I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to," he added

[APEX @ApexUSA1]

A Cabinet Office spokesperson did not comment on whether the latest model had been withheld from the AISI - instead saying it continues to collaborate closely with industry partners, including Anthropic, to make models safer

[APEX @ApexUSA1]

These will soon be superhuman systems that can hack anything, revolutionise any field overnight, and acquire real power and resources.

[APEX @ApexUSA1]

AI agents that break free and hack into other systems are only trying to make us happy.

[APEX @ApexUSA1]

People talk about humanoid robots replacing warehouse workers.

Much bigger possibility:

robots eventually become the default labor force for environments humans were never designed to survive.

Mines.

Oceans.

Disaster zones.

The Moon.

Mars.

[APEX @ApexUSA1]

The next phase of AI will not happen entirely inside a browser.

Claude will move into robots.

AI will move into cars.

Autonomous systems will move into orbit.

Intelligence is leaving the screen.

[APEX @ApexUSA1]

The biggest mistake people make with robots is judging them by how awkward they look today.

The robot is not the important part.

The intelligence controlling it is.

Give something like Claude a body and the trajectory changes completely.