jparishy
re: regular people and their money, I will say, something about the bigger picture just feels off. I look at the graphs you linked to, and do not see it reflected in the people around me. I don't mean to be dismissive of the data, more that I feel like I am missing something, or it is, if that makes sense?
To me, the entertainment industry suffers from a lot of the same problems in most other industries that chase "efficiency", in terms of resources used and profit made. Economies and industries and labor are interdependent and more "alive" than balance sheets and bank accounts.
For the film industry, when you start moving things away from LA to "save money", it has knock on effects that happen over long periods of time. Multiple Quarterly Earnings Reports long, if you will. Local businesses that depend on the film industry close due to less work, so then there is less expertise available. Less expertise means it can command high prices. Suddenly it's too expensive to film in LA altogether. So the rest of the jobs leave too.
This pattern seems super common, not just in the film industry.
It is a narrative that interweaves with AI, too. AI is supposedly an efficiency boom. In the macroeconomy, who is the "efficiency" we are chasing for, who benefits? Who is getting the money that was "saved"? Regular people do not have more more money or time. The "efficiency" is for people who are already very rich, to get richer. Is that worth the price we're paying? I know my answer. The article gets close to it by mentioning the consolidation in Hollywood, but I do not see how we fix this without enforcing our antitrust laws. It is embarrassing what we let rich people do in the name of stock price go up.
One of my side projects is a game, and I've made it so agents can spawn off their own world shard to play in. There's a little Chrome-driven agent harness that can grab low res screenshots while it speaks directly to the game over the wire to control things.
I've had Luna play it and it can drive around (you play as a bike) but it did struggle to play one of the mini-games (Disc Golf). My idea was that model trainers could run it an RL loop using the harness to get feedback data and screenshots to use, then reset the game world as needed during the process.
A smart model capable of multi-modal inference on screenshots/game context at 30-60fps could be pretty interesting, and I'm sure it's coming. No idea how much it would generalize but would be fun to see done.
I do think 95 and 676 will be in the game more accurately eventually. But it will match the style of the rest of the game. It's a work in progress now but not good enough to ship yet. Love that you said bizarro, feels like i'm on the right track
Similar concept, but more abstract in terms of art style. I'm pretty close to having the foundation I wanted before adding quests on top; telling a story is the ultimate goal.
Building on top of GIS data is super fun though. Especially since the barrier to entry is so low these days with LLMs. I encourage everyone to give it a shot if your local town/city publishes some data. No need to understand coordinate systems for example, which is nice.
For the game, there's a button to get in without an account so you can play around, but to get the fun mods you will need to play a bit and figure out the trash pick up/scrap at dump mechanic in the game.
(*might be a bit of lag; server needs more ram but on vacation rn!)
However, most tools available to police are designed to be responsive; they help catch a criminal that has already done something bad. This is not crime prevention. Nothing was prevented. Anyone living in an American city knows these cameras don't prevent much.
It is also (unfortunately) a very vendor driven industry. IACP is a big deal each year, and Flock / Axon / Motorola go all out on spending to influence decision makers.
Talking to communities, in my experience, they ask for crime prevention. People want the actual incident to never have occurred in the first place. When vendors go around saying they are solving crime once and for all, but all they deliver is surveillance, it builds distrust. It feels like lying, while the companies themselves get rich on public money.
Then they go on huge PR campaigns to tell the public actually what they're feelings is wrong and they should accept the surveillance. It's silly.
In the policing communications world, people often say trust is built in drops, and lost in buckets.
Another thought that came from the same post is that, insofar as we see LLMs as human-style intelligence, they're more like stream of consciousness devices. Essentially incessant talking and buying enough time until you get to a usable answer. I think I associate some subset of intelligence with what you don't say, which is impossible with the SOC-style outputs, so this is something I think about a fair bit.
What could maybe differentiate current gen models from next gen is the ability to call tools modeled within the layers themselves, not externally. I think as far as I understand it, model trainers expect the model to do this itself in a way we don't understand or control, like a version of the bitter lesson. But I posit we can model many determinate tools as NNs themselves and figure out how to get the internal states of the LLM to make use of them during inference, e.g. calculators, indexes, citations. Just an enthusiast though, so grain of salt and all.
for my version I leaned arcade: during the race you can jump higher, it gives in game direction hints, and the whole track has a ribbon on it so it's obvious where to go. however, every hour the track changes to a different stretch of Philly roads and it can definitely highlight your point.
The art direction and sense of scale are important, and successful games seem to be quite good at tricking you into thinking the visual scope is bigger than it is, while the smaller/tighter areas are easier to navigate in-game. The real world is surprisingly huge.
(I don't want to spam my game but you can find it in my recent comment history)
Tangential, but I'm also a little obsessed with taking municipal data and turning it into something else. Lately I've been building an online open world game built on Philadelphia's terrain. It similarly uses a Philly LIDAR scan, the land use maps, the street data, building lots, and a few other sources to make the game map.
There's a signup but it doesn't validate or anything I just haven't gotten around to making a demo mode. Any fake email will do: https://cityrider.jpsmaps.com/. The general idea is to pick up trash around Philly to power/build other fun things like your jet pack and ramps. Buy properties and customize the facades. Run from Gritty. It's been a blast, and while not conservation, I believe what is effectively a littering PSA to be a worthy output of my time ha.
You can also click on many trees to see what they are, as most are from the Philly Tree Inventory (though more are added to fill out forest areas for style reasons).
companies must do this, 'cause if they don't then their competition will (i.e. the pressure)
of course, we can collectively decide to equally value labor and profit, as a symbiotic relationship that incentivizes long term prosperity. but where's the memes in that
As a side benefit, when you do this enough, the pendulum that goes over the middle line for any of these arbitrary-but-improves-clicks division builds momentum until it hits the extremes. On either side-- it doesn't matter, cause it will swing back just as hard, again and again.
As a side benefit the back and forth of the pendulum is very distracting to the public so we do not pay attention to who is pushing it. Billions of collective hours spent fighting with no progress except for the wallets of rich ppl.
It almost feels like a conspiracy but I think it's just the direct, natural result of the vice driven economy we have these days
The people in power do not want to lose control, but clearly have no idea how to manage the scale of what has been built. There's an American leadership crisis going on right now that is hard to ignore, both in public and private life.
IMO there's two economies, maybe divided by those who participate in the stock market and those who don't. We, Americans, have largely given up trying to improve the lives of people not in the first group. Economies are living, breathing entities and we're just grinding poorer people for fuel so richer people can have another house, another boat, another company. A lot of regular joes are really stressed out about paying rent. The loss in faith is warranted.
In my experience as a human, the more you know about a subject, or even the more you have simply seen content about it, the easier it is to ramble on about it convincingly. It's like a mirroring skill, and it does not actually mean you understand what you're saying.
LLMs seem to do the same thing, I think. At scale this is widely useful, though, I am not discounting it. Just think it's an order of magnitude below what's possible and all this talk of existing stream-of-consciousness-like LLMs creating AGI seems like a miss