Nobody wants a biased AI like xAI. If I want a biased opinion, I can just ask my neighbor. A good AI feels like the collective knowledge of humanity at my fingertips, not the random ramblings of a lonely old man.
Elon Musk pushes out more xAI founders as AI coding effort falters
791–800 of 889 posts
Re: Elon Musk pushes out more xAI founders as AI coding effort falters
#792Re: Elon Musk pushes out more xAI founders as AI coding effort falters
#793Earlier quoted context omitted.
> I cannot understand at all somebody, who is not Elon, liking/preferring "grokipedia" as idea or implementation. Really? Have you used AI to write documentation for software? Or used AI to generate deep research reports by scouring the internet? Because, while both can have some issues (but so do humans), AI already does extremely well at both those tasks (multiple models do, look at the various labs' Deep Research…
> Have you used AI to write documentation for software? Hi. I have edited AI-generated first drafts of documentation -- in the last few months, so we are not talking about old and moldy models -- and describing the performance as "extremely well" is exceedingly generous. Large language models write documentation the same way they do all tasks, i.e., through statistical computation of the most likely output. So, in no…
That's not at all been my experience. My experience has been one of constant amazement (and still surprise) when it catches nuances in behavior from just reading the code.
I'm sure there are many variables across our experiences. But I know I'm not imagining what I'm seeing, so I'm bullish on the idea of an AI-curated encyclopedia, whether Elon Musk is involved or not.
Re: Elon Musk pushes out more xAI founders as AI coding effort falters
#794Grok 4.20-beta1 scores above GPT-5.4-high and just behind Opus 4.6 on LMArena for Text https://arena.ai/leaderboard I guess for coding if you’re not first you’re last, but this is damn impressive considering. It looked like they pulled the coding model from the benchmarks, but it was similar.
Accroding to https://artificialanalysis.ai, it's around Gemini Flash 3, or some of the Chinese open weight models, like GLM 5.
For all the money burned, I am not impressed. Why would I use Mecha Hitler for almost double the cost of Gemini Flash 3?
Re: Elon Musk pushes out more xAI founders as AI coding effort falters
#795It feels like xAI is perpetually playing catch-up. They haven't quite committed enough to a novel direction relative to anthropic or OAI, what's described in the OP seems symptomatic of a lack of differentiation. If you spend all your time judging yourself relative to the incumbents, there will be no time left over to innovate. The leash is too tight!
For a brief moment, they were the top performers in benchmarks with the release of Grok 4.
Then they suddenly fired tons of people. Elon does not understand the market and the competition. You can't run a frontier AI lab like any old VC slop company.
Re: Elon Musk pushes out more xAI founders as AI coding effort falters
#796Used Grok for the first time, in a Tesla, and for that purpose it actually made a lot of sense. It’s very well-integrated into the car’s systems and communication style while driving tends to be very tweet-esque. I think this is the niche they should lean into more (live assistant, e.g. Jarvis type stuff) and leave the more agentic niche to folks like Anthropic. Maybe even delegate more difficult or background tasks…
What's the difference between Jarvis and agentic?
Is this a serious question? In my view a live assistant differs from an agentic model in that it is a conversational interface, interconnected with live data feeds and able to exert some control over the physical environment. Can agents do some of those things? Sure. But this isn’t the primary intended application for agentic tech, which focuses on running longer tasks unattended. And yes, Jarvis could make use of background tasks himself, and I think that’s a major part of his value. But it doesn’t have to be Jarvis actually performing those tasks, just directing them. Let the agents do the agentic things, and let the more conversational models interface with the humans.
Re: Elon Musk pushes out more xAI founders as AI coding effort falters
#797Earlier quoted context omitted.
> What is the benefit of "moving compute to space"? I’ll bite. It’s cheaper and quicker to permit a launch than permit, zone and interconnect a datacenter. And solar panels in space don’t need glass cladding, which makes them cheaper to make and lift. The downside is launch cost. But there is a breakeven between these factors that seems to have most of its error bars within Starship’s target. (By my math, around $35/…
> It’s cheaper and quicker to permit a launch than permit, zone and interconnect a datacenter There's plenty of empty land sufficiently far from cities and not being used for anything else and that shouldn't have permitting or zoning problems. For interconnect do that via satellite.
> plenty of empty land sufficiently far from cities
Which means interconnect permitting.
> For interconnect do that via satellite
As in power.
Re: Elon Musk pushes out more xAI founders as AI coding effort falters
#798Earlier quoted context omitted.
No, when did I say that? That’s impossible for anything of the size of Wikipedia. I was suggesting that Elon Musk, a man who has donated hundreds of millions to Trump and other Republican causes, who has numerous financial conflicts of interest, and who has publicly lied numerous times, is never going to produce a more unbiased and factual encyclopedia than Wikipedia. Especially when his effort to do so is essentiall…
Right, and Wikipedia leadership is free of any conflicts of interest: During and after her Wikimedia role, Maher drew fire for statements perceived as rejecting objective truth or Wikipedia’s traditional “free and open” model: • In a 2021 TED Talk, she described reverence for truth as potentially a “distraction” hindering common ground. • She called Wikipedia’s free-and-open ethos a “white male Westernized construct”…
But she’s not the one writing and editing the wiki pages. It’s open, and there are open discussions as well. What’s it matter what she thinks?
Why would I ever trust an encyclopedia totally controlled by one megalomaniac over one that I myself can contribute to?
Re: Elon Musk pushes out more xAI founders as AI coding effort falters
#799Earlier quoted context omitted.
> What is the benefit of "moving compute to space"? I’ll bite. It’s cheaper and quicker to permit a launch than permit, zone and interconnect a datacenter. And solar panels in space don’t need glass cladding, which makes them cheaper to make and lift. The downside is launch cost. But there is a breakeven between these factors that seems to have most of its error bars within Starship’s target. (By my math, around $35/…
The capacity of a single datacenter would require thousands of launches to get the equipment into space. I don’t believe for a second that this would be easier in any way. Cooling and bandwidth are also completely unsolved for compute on a useful scale.
> capacity of a single datacenter would require thousands of launches to get the equipment into space
But that equipment starts generating compute as soon as it’s up. This dramatically increases the capital efficiency of the venture. (Though space launch is still ultimately capital intense. Lower rates go, the more attractive it becomes.)
> Cooling and bandwidth are also completely unsolved
Quite wrong. (Though I was surprised by this, too.) ISS-style radiators (14 kg/kW) require Starship’s most optimistic launch cadences to make economic. But sub 10 kg/kw, which is closer to ISS heritage than any of the newer stuff, lets $100/kg to LEO work under most circumstances. Drop it to 6 kg/kW and even Falcon 9 becomes viable for low costs of capital (Bandwidth is a problem, but an engineering one. (And one Starlink is working on with laser backhaul.)
Re: Elon Musk pushes out more xAI founders as AI coding effort falters
#800Earlier quoted context omitted.
> What is the benefit of "moving compute to space"? I’ll bite. It’s cheaper and quicker to permit a launch than permit, zone and interconnect a datacenter. And solar panels in space don’t need glass cladding, which makes them cheaper to make and lift. The downside is launch cost. But there is a breakeven between these factors that seems to have most of its error bars within Starship’s target. (By my math, around $35/…
What about maintenance? I’d naively assume that’s the killer.
> What about maintenance?
Simply put, you don’t. Your DC is launched into its graveyard. If a chip burns out it burns out—maybe rack design is a bit more redundant to keep failures as independent as possible.
Maybe at some point repair is a valid optimization. But it’s not necessary for an MVP, namely, one that is competitive against 3 to 5-year terrestrial delays and sub-10% costs of capital for such projects. That’s what has surprised me.