HNHacker News
TopNewBestAskShowJobs

pj_mukh

4,439 karma · joined December 4, 2013

Building Spatial/Physical AI for over 15 years.
submissionscomments
pj_mukh··on Show HN: Open source framework OpenAI uses for Advanced Voice
Can I currently put a VAD module in the pipeline and only send audio when there is an active conversation? Feel like just that would solve the problem?
pj_mukh··on It's Time to Stop Taking Sam Altman at His Word
Except the article makes none of these points. The article is saying:

a) "the technology is overhyped", based on some meaningless subjective criteria, if you think a technology is overhyped, don't invest your money or time in it. No one's forcing you.

b) "child abuse problems are more important", with a link to an article that clearly specifies that the child abuse problems have nothing to do with OpenAI.

c) "it uses too much energy and water". OpenAI is paying fair market price for that energy and what's more the infrastructure companies are using those profits to start making massive investments in alternative energy [1]. So if everything about this AI boom fails what we'll be left with is a massive amount of abundant renewable energy (the horror!)

Probably the laziest conjecture I have endured from The Atlantic.

[1]: https://www.cbc.ca/news/canada/calgary/artificial-intelligen...

pj_mukh··on Show HN: Open source framework OpenAI uses for Advanced Voice
Super cool! Didn't realize OpenAI is just using LiveKit.

Does the pricing breakdown to be the same as having a OpenAI Advanced Voice socket open the whole time? It's like $9/hr!

It would be theoretically cheaper to use this without keeping the advanced voice socket open the whole time and just use the GPT4o streaming service [1] for whenever inference is needed (pay per token) and use livekits other components to do the rest (TTS, VAD etc.).

What's the trade off here?

[1]: https://platform.openai.com/docs/api-reference/streaming

pj_mukh··on YC criticized for backing AI startup that simply cloned another AI startup
Agreed. Looks like the founder agrees too[1], that was a pointless waste of time.

[1]: https://x.com/CodeFryingPan/status/1840831339337302204

pj_mukh··on YC criticized for backing AI startup that simply cloned another AI startup
AFAICT, the attribution's been there since the beginning on their GitHub? (see: About section)[1].

I agree adding their own license was hamfisted, but honestly, if I'm funding a company I hope they would spend less than <$1 on legal and licensing initially. The first order of business is almost always proving whether your business should even exist.

https://github.com/trypear/pearai-app?tab=readme-ov-file

pj_mukh··on Y Combinator Traded Prestige for Growth
Yea it does! [1].

[1]: https://jaredheyman.medium.com/on-the-176-annual-return-of-a...

pj_mukh··on Y Combinator Traded Prestige for Growth
Did they? YC regularly funded competitors. This was always true. Fundamentally, YC is betting on founders. They're optimizing for founders who can move fast enough to find a moat before they die, that's it.

They fund competitors because most YC companies usually pivot out anyway.

pj_mukh··on Y Combinator Traded Prestige for Growth
This is an indictment of Twitter tbh. Not YC. A lot (most?) YC founders aren't even on Twitter.
pj_mukh··on Y Combinator Traded Prestige for Growth
"did some clunky work on it. That was their entry to YC"

For more context, YC doesn't judge your code, never has. It was never a code quality competition. Orthogonally, they do judge the results (user metrics).

pj_mukh··on Y Combinator Traded Prestige for Growth
"The issue isn't that PearAI did something illegal—it's that they got funded by YC with nothing more than a codebase copied from another YC-backed company. This shows that (1) YC is willing to fund just about anything, (2) they’re not doing any real due diligence, and (3) they don't particularly care about their existing portfolio companies."

This shows a fundamental misunderstanding on how YC functions. YCombinator was never a test of how good (or unique) your code was. At its core it was a filter of people, people who can work well together and people who can build something useful. That's it. You can read more about this straight from one of the founders [1]. The fact that you used open-source code (within legal bounds) to get their quicker just shows your resourcefulness, something YC actually optimizes for.

More often than not, "good people" tend to be domain experts sometimes really good and unique coders but that, to me, was always a byproduct of the search pattern.

You can obviously disagree with this methodology, but it has worked pretty well.

[1]: https://foundersatwork.posthaven.com/the-social-radar-what-i...

pj_mukh··on Gavin Newsom vetoes SB 1047
Or used a model someone open sourced after spending $100M+ on its training?

Like if I’m a startup reliant on open-source models I realize I don’t need liability and extra safety precautions but I didn’t hear any guarantees that this wouldn't turn off Meta from releasing their models to me if my business was in California?

I never heard any clarifications from the Pro groups about that

pj_mukh··on OpenAI to become for-profit company
“the winners get space ships and the losers die, they'd call you crazy”

Was the original quote. I just showed that you could get rid of all rockets ever made and it would be a rounding error for any funds used to save the “losers” from dying.

How is that a non-sequitur?

pj_mukh··on OpenAI to become for-profit company
My point was every other problem would be made worse by a billionaire pushing his/her money in there. Everyone is a couple of billions of money funneled away from becoming the next George Soros.

If you don’t think NIMBYism and degrowth is a problem in your country yet, just give it a couple of years. It just hit England, you’re next. No billionaire can save you.

pj_mukh··on OpenAI to become for-profit company
Literally every problem I mentioned is at its worst state possible. People with millions and billions simply waiting to buy materials or get land approvals. It’s a well know intractable problem [1] and really the crux of the problem.

If just these problems could be solved the state has more than enough funds to house everyone. What billionaires do would be wholly irrelevant (like it is now)

[1] https://www.constructiondive.com/news/construction-materials...

pj_mukh··on OpenAI to become for-profit company
Yes this has basically been the modus operandi of the gates foundation and it took them 10 years to make a dent on Malaria. They still have no clue how to “efficiently” reduce famines.

They won’t touch American housing problems with a 10ft pole. That should tell you something.

Go to Berkeley, tell them a Billionaire wants to build housing for the homeless in their neighborhood. See what happens.

It’s a hard pill to swallow but the best thing billionaires can do is let us tax them and then butt out go fly rockets. The political problems is upto the rest of us.

pj_mukh··on OpenAI to become for-profit company
I didn’t say macro scale, I said Los Angeles that’s the problem.

Pretty much everything required to build housing, wood, labor, pre-approved land is in a massive shortage that we can’t spend money to fix.

So more money to simple pump demand for all those things will have a massive inflationary impact.

pj_mukh··on OpenAI to become for-profit company
"An uptick in construction work will not be an inflation balloon. "

There's a massive shortage in construction workers [1], so yes there will be? The few construction workers we do have can demand higher wages (yay!) but then will they be outbidding other mid-income folks for housing with those increased wages? Sounds like an inflation spiral to me.

My statement wasn't against social housing, I love social housing. We just haven't cracked the code in scaling housing (and subsequent maintenance) yet. And the problem is about 80% political will, billionaire cash is useless here.

[1]: https://www.abc.org/News-Media/News-Releases/abc-2024-constr...

pj_mukh··on OpenAI to become for-profit company
"where the winners get space ships and the losers die,"

The Social Security budget is $1.4 Trillion, just the federal welfare budget is >$1Trillion (not including state budgets), and then there's medicare. Meanwhile, the NASA budget is <$25B (with SpaceX's operating budget and profits being a fraction of that)

I wish we lived in that simple of a world. But we don't.

pj_mukh··on OpenAI to become for-profit company
"Feed the hungry. House the homeless"

Funnel $10B in housing to Los Angeles and you'll build less than 100 units of housing, because the inflationary push of that money would balloon the cost of per unit housing. I don't want to imagine the effect of that on middle class housing.

Funnel $10B of food to xyz famine region and you've undercut local farmers for generations. Happens all the time [1]. And that's assuming you can get the aid past local corruption.

These problems aren't as simple as people assume, and I'm low-key happy young naive Billionaire's are avoiding these issues instead of trying to throw their weight around.

FWIW: Sam's already funneled a bunch of money into green energy production[2].

[1]: https://haitisolidarity.net/in-the-news/how-the-united-state...

[2]: https://www.cnbc.com/2024/09/25/sam-altman-backed-nuclear-st...

pj_mukh··on In the Shack with Robert Caro
Can you jerry-rig something to code on it?
pj_mukh··on In the Shack with Robert Caro
"A couple of bookcases, a plywood work surface, corkboard with outlines tacked up, an old brass lamp, an underworked laptop for emails, a Smith-Corona typewriter. "

I wonder if software engineers would pay for some software (or dedicated hardware) to turn their laptops into a coding typewriter with maybe just a GPT connection to probe documentation and whatever tunnels they need to work on internet deployments.

I'd love to experiment with something like this.

pj_mukh··on OpenAI to become for-profit company
I would agree with you that a space agency is also useful (maybe more useful some days of the week). Sam disagrees and thinks he can do better without a non-profit board now. I'm glad we live in a world where he gets to try and we get to tax him and his employees to do other things we consider useful.
pj_mukh··on OpenAI to become for-profit company
What do you do with a a lot of money past a point? A corporate controlled AGI being just a stop on the way to build another private space agency seems like a...letdown.
pj_mukh··on OpenAI to become for-profit company
Occam's razor: I think Sam's personal narrative is the correct one. He built a non-profit that took off in a way that he didn't expect it and now a for-profit is the best way to run the lightning they've caught.

In terms of profit, AFAICT, Sam doesn't have designs on building extra large yachts and his own space agency but what he wants is to be the one at the stead of building what he considers is world-changing tech. One could rationally call this power-hungry but one could also rationally call this just helicopter parenting of a tech you've helped built. And for that a for-profit that is allowed to maximize profits to re-invest in the tech is the optimal setup (esp if all the competitors are doing the same)

Is this a different org than when it started? Yes. Was this a dupe from the beginning? I don't think so.

"But why can't he have a more worldly-aligned board looking over his shoulder?"

Because we live in California and have a bad taste for governance by committee or worse: governance by constant non-representative democracy (see: Housing).

If this now completely comes off the wheels, I still think Congressional action can be a stopgap, but atleast for now, this restructure makes sense to me.

pj_mukh··on Orion, our first true augmented reality glasses
I think the main question will be if Meta can run the glasses under a different business model than ads/personalization. I know Meta feels like they missed out on owning the hardware ecosystem with mobile, but a big part of why Apple's product work is because their business model is not ads.

If they can fix their incentives there it will address 80% of the concerns in this thread.

Counterintuitively I think a passive screen affixed to your face will reduce your reliance on all other screens. Glasses aren't a great form factor for scrolling feeds, but they are much better at connecting you to your physical world (which phones are the worst at). That plus just the posture improvements [1] that glasses can provide over the phone may make this a winner.

[1]: https://www.today.com/health/texting-neck-how-hunching-over-...

pj_mukh··on Orion, our first true augmented reality glasses
Right, though publicly available for $99/mo.

With Orion, no devkits publicly available either.

pj_mukh··on Orion, our first true augmented reality glasses
FWIW, Orion is not for sale.

Exciting work! It seems like the main problems are now in miniaturizing electronics (and not optics) into a Ray-Ban form factor? Super cool.

pj_mukh··on Launch HN: Modern Realty (YC S24) – AI Real Estate Agent for Home Buyers
So amazing! Strongly considering using this for my next purchase. Two quick questions:

a) How do offer + contingencies work (contingent on financing, home inspection etc.), will a human immediately have to get involved when any contingencies are involved?

b) Your service will text the sellers agent, and get me the lockbox number so I can tour a property myself without having to schedule with a buyers agent? Just that service alone would be amazing

pj_mukh··on Bill limits smartphone use in California schools
Man, so many of Californias problems is just:

“vocal minority is very loud, everyone else disagrees but are too busy to come yell at electeds, so we can’t fix this obvious problem. Done. Problem festers”

pj_mukh··on 'I Don't Want to Die.' He needed mental health care. He found a ghost network
In cases of public healthcare the problem is almost always a shortage, doctor shortage, bed shortage, nurse shortage, tech shortage etc.

The perfect system is a public healthcare with true abundance.

← PreviousPage 21 of 34Next →