poslathian

Seconded, great share

Really!? Glm5.3 is my daily driver and I feel im having the most productive experience with agentic collaborations so far, by a lot. Using pi with tons of custom extensions, that to be fair I developed since making the jump off of codex and claude about 12 weeks ago. I primarily do not write code for a living. I do a lot of modeling and commercial analysis and a lot of math (related to differentiable simulation)

I’ve heard this about mechanical electronic signage used for team dashboards 30-40 years ago.

You’d sit in the office with your back to it and the clicky clacky sounds would be going every so often when the content updated. but people would know the specific clicky clacky sound that required them to turn around and read the sign based on their job function.

I’ve also heard of day traders mapping stock tickers or other data into sound and having headphones on all day with it

Interesting point. I think there might be something special about sag given the way Hollywood does business. Everyone switches employers constantly as projects spin up and wind down. Having a mostly permanent role in an organization is super unusual - and I wonder if these roles (like a 20 year soaps gig or the simpson’s) have different stats on working on/off the card.

Nice, very LTL

The bibliography is more important, imo, than the peer review. I get the most use of arxiv surfing references and citations.

This take is oversimplified to where it’s literally scary - heres an honest book about the Rothchilds Bankers and Zionists with the additional context I hope you incorporate.

https://ia601309.us.archive.org/20/items/historyDEEPWEB/The%...

Sandberg built the company that polluted everything with this cynical influencer economy. Schmidt built the company that turned users into the product.

Oh yeah, his decision here updated my mental model a lot

I’m lucky to have more than my share of really exceptional programmers to hang out with and they all say the same thing: “I haven’t been writing code for months and don’t expect to again”

This is a way different sentiment than “programmers aren’t needed anymore” - I’m just seeing ambition, motivation, and fun go up in lockstep.

I first heard this in November and slowly one by one it’s everyone whose opinion I respect.

FWIW the other popular topic is how abysmally stupid and limited these amazing tools continue to be, despite also being magic.

Oh and that none of us have gotten token maxxing to succeed, despite lots of trying.

What if you are going to spend 5400 either way, you go all agent or get an apprentice and an agent for them too.

This is the freshest take I have ever seen. Kudos, so cool.

My professional career has been 20 years of physical ai research and now industry. My dad insisted I take corporate accounting in undergrad.

It’s a skill I use every single day, double entry bookkeeping is one humanities great inventions and deeply related to conservation laws utility in various other areas. Could not have built my business without having gotten so into this Topic.

Pushing 20 years in industry and this is the best take I’ve heard

as I understand, Bell Labs mandate was to improve the network, which had tons of great threads to pull on: plastics for handsets, transistors for amplification, information theory for capacity on fixed copper.

Google and Meta are ads businesses with a lot less surface area for such a mandate to have similar impact and, frankly, exciting projects people want to do.

Meanwhile they still have tons of cash so, why not, throw money at solving Atari or other shiny programs.

Also, for cultural reasons, there’s been a huge shift to expensive monolithic “moonshot programs” whose expenses need on-demand progress to justify and are simply slower and way less innovative.

3 passionate designers hiding deep inside Apple can side hustle up the key gestures that make multi touch baked enough to see a path to an iPhone - long before iPhone was any sort endgame direction they were being managed to.

Innovation thrives on lots of small teams mostly failing in the search for something worth doubling down on.

Googles et al have a new approach - aim for the moon, budget and staff for the moon, then burn cash while no one ever really polished up the fundamental enabling pieces in hindsight they needed to succeed

They are differentiable which allows for image based rendering via solving the inverse of the rendering function via gradient decent

Absolutely!

These techniques are the key unlocks to robustifying AI and creating certifiable trust in their behavior.

Starting with pre-deep neural network era stuff like LQR-RRT trees, to the hot topic today of contraction theory, and control barrier certificates in autonomous vehicles

We always called this “monetizing the brand” and it’s been annoying me since at least when Sperry when private equity and the shoes stopped being multi-year daily drivers

Is it possible that regardless of what they pay they think Anthropic is negative margin on it?

This true - and git was not a moving target. AI core tech has certainly slowed down but still moving fast enough to make hard won lessons worthless and investing in learning them questionable.

After becoming familiar with the reality of the cost inflation of (in my case local government real estate) development projects vs private I chalked it up to graft, incentives, and mismanagement.

Actually your comment is probably more correct - adds a whole step to move the wallet. Misaligned incentives and mismanagement are probably more equal across public/private than we like to believe

I love this quote. mankind having been to space already, the rockets are the sideshow to the way they designed and grew an org that delivered them along with a great business, starting from an amount of capital loads of nobodies have had but failed to do anything interesting with.

Is it odd or is it kinda central to the point?

It’s exceptional that people centrally organized a huge amount of effort and resources towards something imagined by countless humans since prehistory, was far from being a sure thing, had no possibility of revenue and only indirect value, planned and executed a full decade toward a single objective, and succeeded in a single moment shared by almost everyone with a television.

Arpanet, the transcontinental railroad, the pyramids…amazing still but lacked the 0 to 1 all at once factor. Starship is inspiring and also not a moonshot.

Given the circumstances and the relatively low dollars involved, it would be interesting to see the experiment: $10B darpa program to establish a scalable fab ecosystem in 5 years via consortium.

This was how the internet was created, darpa stitched together dozens of performers to get the key ingredients (eg bbn gateways, academic subnets, experimental applications, protocol research.

They even led the last ditch marketing Hail Mary after years of no-one caring about the program besides the zillions of engineers from all around building it by organizing a press day in a hotel ballroom for a demo day.

As a taxpayer I’d strongly support 5B/.1% of the fed budget for a few years just to learn what happens in the attempt.

It’s super debatable whether or not DARPA has done more for creating enterprise value in the US tech sector than sand hill.

At the least, without darpa the whole Bay Area machine would not exist today, so it’s at least necessary if not sufficient.

Not just darpa but nih, nsf, doe, onr, arl, nasa, and the national labs are definitively necessary causal dependencies on of every company and industry driving US national pride and all of the most valuable companies.

Even if the firm never takes a grant, their talent, supply chain, and component pipelines all depend on these grantor agencies thoughtfully allocating taxpayer capital at the national level.

This comment gives me mixed feelings and some nostalgia for when our company was < 100 people and one of the core software teams was called “meow” - today we call it human robot interaction.

For those wondering how to connect PID to the foundations of AI. https://en.m.wikipedia.org/wiki/Cybernetics

I like this metaphor about electric lighting. However, having lived in two ~1850 houses, they sure look and function a lot like they did before electricity, despite nearly every element having been “disrupted” by electricity and all the rest.

I’ve worked in this field for 20+ years and as far as I can tell the only consistent colloquial definition of AI is “things lay people are surprised a computer can do right now”

The way this is rolled out is super weird. Ultra bold - shocking - claims by a credible researcher but zero details amidst tons of documentation and even code that would hint at how these two universes (model based simulation and model free generative models) are being unified beyond a lot of references to differentiable simulation…

Super exciting!