Elon Musk’s Weekly X Files: What He’s Really Saying Online

elon musk just admitted his ai grok isn't the preferred choice for some tasks compared to its rival claude last week the talk was all about ai upgrades this week musk spent his time explaining what his ai can't do on september 12 he replied to a user comparing the two ais saying grok is not yet a preferred choice in this arena not sure how to feel about that that post got nearly 200,000 views it was a moment of public concession and it wasn't isolated the day before he announced that the next version grock 4.7 needed more time in the oven his post which got 2,000,000 views was specific he said the model gives up on hard

tasks that it can do too early and isn't yet sufficiently rigorous in checking its work this is not the usual tech ceo playbook this is moving slow and telling everyone in public exactly what's broken now while he's being candid about grock's flaws the hype for everything else is turned all the way up it's a study in contrasts for example he used grock to summarize a talk by the spacex cfo at goldman sachs the summary which he posted to 2,600,000 viewers highlighted spacex's core model of vertical integration that is owning everything from the raw metal to the engines the software the launch pads and the satellites themselves so here you have the flawed ai being used

as a promotional tool for the rocket company that has mastered its supply chain it's a clever way to show grok is useful even if it's not perfect while reminding you of his other successes then there's tesla on september 8 he posted that they have redesigned production for the cybercap his claim it's over five times faster than conventional automotive manufacturing that post got 7,000,000 views it's a direct shot at the entire global auto industry and of course there's x itself he made sure to post that it is the global town square and that it has surged to become the number one news app in nearly all european countries so what does it all add up to you have these very public very specific admissions of failure with grok and right alongside them you have these monumental claims about manufacturing space and media and then he zooms out even further he posted a prediction viewed almost 5,000,000 times that ai and robots will more than double the global economy in less than ten years it's a classic

move when the details on the ground are messy you point to the horizon this isn't just random posting it's a calculated performance of transparency he's showing you the engine being built acknowledging the loose bolts so you'll believe him when he says it's gonna fly to mars by admitting the small failures with grok he's buying credibility for the gigantic claims about tesla spacex and the future of the world economy he's trading short term ai pride for your long term trust in his vision of course there's always the background noise he spent part of the week swatting away critics a four hour documentary was dismissed as a hit piece that misses and makes the fatal sin of

being boring for four hours allegations from a media personality were met with a single laughing emoji he posted a strange context free link referencing a real professor at carnegie mellon which stirred up controversy and then just vanished it's all part of the same broadcast the signal and the noise all playing at once this week sets up a new question the story is no longer just whether grok will get better it's whether this entire strategy of radical public honesty actually works in the long run he's not just building an ai in the open he's building a case for why you should trust him even when his products fail