if codex can write the code,
what are you actually hiring the engineer for?
debugging the weird edge case?
making the product decisions?
knowing which 80% to skip?
most interview answers still sound like 2021.
what's the real answer now?
@VadimStrizheus The next will be ability to run these open source model in slim hardware.
Because cloud Frontier doesn't look like the future of personalization.
Affordability also matters!
@elonmusk What will humans do in the meantime wait and watch? Because i assume even bots will be made by another bots? Are we making humans useless?
Cause not everyone is engineer to get paid from this ?
@JonBuildsHQ@mostly_error Yes, made on extension locally and got flagged by system, had to remove it and appeal for my account.
in my case i was not even generating replies. just viral worthy post!
But seems extension like this will not work against X's current system.
@tankots@WisprFlow I have not used it but would love to know how is this different from the in-built voice typing in windows, that also serves the same purpose, especially from the reviews below it is showing the same issues, that i encounter in the in-built feature.
People keep saying “$60B acquisition” like someone wired $60B.
No. This is private SpaceX stock being treated like money.
If SpaceX valuation crashes tomorrow, Cursor holders are left holding paper, not cash.
Amazing move by Elon, but pretending this is the same as a cash deal is absurd.
SpaceX buying Cursor with stock is the actual story here.
No $60B cash going out. Just SpaceX equity turning into money.
Once your private company has a monster valuation, you can basically go shopping with paper and pick up real companies.
Elon may have found the ultimate acquisition cheat code.
So much of non-sensical information, first you can't run ai models on 10k INR smartphone, you need a flagship phone and highly quantized model.
Second cheap AI and useful AI are 2 different things, there are lot of open source models that you can run but they are often not useful for most of the use cases except text generation.
The heavy ones like kimi, glm needs servers and if you wanna run it on local then you need to have a high-end GPU.
In either case, you cannot implement jio model in AI, the hardware cost is way more unexpected and allocation is another thing, reliance commoditized the already existing stable technology of 4g.
whereas AI is still in developing phase!
This is one of the most delusional AI takes I’ve seen.
Ban frontier models? You’re going to build frontier AI on hardware imported from the US, while telling Indian enterprises not to use the best AI tools to upskill themselves?
There is a fine line between strategic sacrifice and stupidity.
What you’re proposing is stupidity.
Use the best tools, upskill fast, build local capability in parallel. Cutting yourself off doesn’t make you sovereign. It makes you slower.
You are mistaking India’s population for India’s AI market. 1.4B people doesn’t mean 1.4B paying AI users.
“Indian languages first” sounds powerful, but language is becoming a feature, not a moat. OpenAI, Google, Meta and every serious model can localize fast.
The real question is not “can it speak 22 languages?”
The real question is: who is paying, how often, and why won’t they just use the global model that already sits in their workflow?
Especially cost-effectiveness. Global labs have deeper compute access and can subsidize subscriptions at a scale Indian AI companies simply can’t match.
And in India, “Make in India” emotion usually stops at pricing. People won’t use something as their daily driver just because it was made in India. They’ll use what is cheaper, better, and more useful.
@Reuters AI companies love open research when they’re climbing.
Then suddenly everything becomes a “trade secret” when the next guy starts catching up.
Funny how that works.
@elonmusk Getting arrested for a retweet should scare everyone.
You don’t have to agree with the post to see how dangerous that is.
Once people are afraid to share the wrong opinion, free speech is already in trouble.
@elonmusk The craziest part is that today’s “impossible scale” will probably look small in hindsight.
That’s usually how civilization-level infrastructure begins
@elonmusk People still think of SpaceX as a rocket company.
This looks more like the early infrastructure layer for a multiplanetary economy.
The scale won’t be understood until it’s already obvious.
I don’t dismiss the value of technology. Society is clearly changing, and kids will grow up in a world shaped by high-tech tools.
But I also think we’re forgetting something important: childhood is not supposed to be optimized like software.
There is real value in boredom, books, outdoor play, exercise, face-to-face friendships, and learning how to build real relationships without a screen in the middle.
@elonmusk Making the Sun sentient sounds like the final boss of civilization.
But the harder problem is still us: building intelligence faster than we’re building wisdom.
Tip culture in the US is absurd!!
A tip should always be a choice, not a compulsion. I don’t understand why businesses don’t just include the real cost in the product or service and pay their employees better wages.
Expecting customers to pay for the service, and then pay extra so the worker can earn properly, makes no sense.