Grock 3 - funniest conversation so far (according to Grok) grok.com/share/bGVnYWN5…

Sounds like Grok3 is keen to make X better grok.com/share/bGVnYWN5…

I’m genuinely interested if anyone has evidence of using deepseek or similar to predict the timing and size of this current crash Seems to be something it should have been reasonably good at with the right prompting

Ok Sam Altman deep research is really dope but why can’t it put results in a pdf or word document for me? Basic o3 queries manage this with ease. Copying it from the output window for long reports is really annoying and time consuming

AI - there comes a point (possibly v.soon) where humanity can no longer verify if what AI says or 'proves' is true or not - so we will need to trust. So we need to start working on mechanisms of trust as a priority

Go algo

Always worth syncing up with the signs and your own intuition - I feel these things always pre-warn to those that attune

Getting better…

Not sure I'm gunna even need GTP-5 when o3 understands all my slang

Worst decision of the weekend already - VAR is ruining this sport

We need to start a movement to make this goal of the season

Hot take: In a post-AGI world, being kind will outrank being rich or famous

Slott has bought back Fergie time! #LFC #ChampionsLeague

My take - this could be the starting gun for the final Oppenheimer-esque race to ASI. The tariff move is likely a cover story. But watch this line carefully: “Also on November 1st, we will impose Export Controls on any and all critical software.” Until now, both sides needed

Imagine if you find yourself in a race to ASI hit an energy bottleneck and have free energy sitting on a black shelf somewhere

Where is Australia’s version of this? - future proof our economy by dramatically improving productivity - utilise our natural resource base in country reducing reliance on exports - bring tech and AI expertise in Country - our models our infra our rules - Improve national

AI Evolution = Enders Game Convince me otherwise

AI can now run cyber-ops alone. Centralized security is done. This isn’t about patching. It’s the wrong architecture. Self-hosted agents. Local memory. Private compute. No exposed surface. We either fragment the stack or risk loosing it.

Prediction for 2026: AI sovereignty moves from an afterthought to a mainstream demand. Look at what AI already knows about you. As an extension of self, privacy, auditability, and sovereignty must be choices - not defaults decided elsewhere. Not your AI. Not your mind.

Most 'agentic coding' today optimizes execution, not judgment. Spec-first + repair-on-drift works surprisingly well - but it’s still waterfall. The missing piece isn’t better prompts, it’s systems that can pause, ask, and renegotiate goals as understanding changes.