Per Aspera Ad Astra

Things are about to get very weird, very quickly...

🤯 OpenAI's 4o model was groundbreaking for less than 24hrs ...

… now Google DeepMind has hit back with Project Astra ... which (watch till the end) is quietly mind-blowing even in a post-4o world.

"This changes everything" is an old trope but ... I think it's genuinely difficult to understand how fast we're moving these days.

The context windows required for these models to act in this way are so massive ... that it belies a capability that puts much of human cognition to shame.

We are brilliant - as a species - but let's be frank: few of us have either the polymath-breadth of deep knowledge or the zero-shot memory that these models already possess.

Where this goes still depends on our political, cultural and economic will to do good with this tech ... but our propensity to use technological advancement to subjugate whole populations should give us great pause.

If you think I'm being hyperbolic - just look at the way insulin is manufactured in the US under an open-source patent and sold in the US at an exorbitant price, while sugar is pumped into the human food chain and the diabetic support lobby denies that diet can help alleviate type-2 diabetes. That's just one example.

Capitalism is not kind to the base of the pyramid - and when you put this level of cognition into autonomous robots (which you can expect within a few years, if not a few months) we face human obsolesce on a massive scale.

There's nothing to stop this advance so all I can recommend for now is that you make sure you live your life with a purpose that goes beyond your job ... and that you teach your children to do the same.

🖖