I feel like this has a much better chance to succeed than lab-grown meat for two reasons:
1. It's grown using bacteria, which are much faster to reproduce and cultivate than animal cells.
2. The intended products are a homogeneous mixture, which means you don't need to worry as much about texture
From what I've seen, it doesn't seem like AI is very good at keeping track of events in a way that makes a logical narrative.
I tried "The Hobbit" and it started with me having tea with Gandalf the wizard. There was a knock at the front door so I opened it and outside was Gandalf the wizard!
I built my own LLM-based interactive text adventure system last year, but in addition to LLMs, it used an auxiliary database to keep track of where everything was, what state it was in, and so on.
The LLM thus did two things: first, it used a combination of the prompt, the user input, and the recent history of the chat and relevant extracts from the database state, to update the database with a new state; then after updating the database, it used all of the above to synthesize the output response.
A major component of the system was that, in addition to the usual blocks-world subject-property-object representation of the game state, the system also used the database to store textual fragments that encoded things like thoughts, plans, moods and plot points.
Adding programmatic constraints to the database and its updates made a vast difference to the quality of the output by preventing logical absurdity of the sort that has been described above - attempts by the LLM to do absurd things were simply rejected.
Finally, there was a hierarchical element to the story-telling, not just simulation. There is much more to storytelling than simulation, and experimenting with this became the most interesting thing about the whole enterprise; there is plenty of space for future research in this.
The downside was that all this complexity meant it was _extremely_ compute-intensive, and thus very slow as a result. But the gameplay was fascinating. We will no doubt see more things like this in future as LLMs get faster; it's an obvious application for open source software.
Fascinating that you actually went through with an implementation. I’ve been throwing the idea of LLMs somehow having a sat solver built into them (maybe trained to have one as an emergent property), but something like you describe is the next best thing.
I've tried quite some models and so far I've only seen two that are good at keeping continuity: Gemini 2.5 Pro and Sonnet 3.7 (both with reasoning on). These two are also exceptionally good[0] at adding plot twists that aren't in the prompt.
[0]: by "my expectation of LLM" standard, not by the standard of bestseller genre fiction.
A context of a few things characters? No, it definitely does.
Beyond that... Would you forget that the guy you were having tea with was sitting in your living room? Anyone I know would definitely go "how the hell are you outside" not "nice to see you today".
2. Could you have reasonably de-escalated (or were you the one who escalated to get here)?
3. Can you convince a jury (and the cops) that your life was in danger? If there are no witnesses, this is tough. Typically you're allowed to use deadly force only if you fear loss of life/limb. Yes, yes, plenty of cases where juries ruled in favor of the shooter when there was clearly no risk of loss of life/limb.
4. Do you have the relevant insurance to cover your legal defense costs? If not, you'll likely make a plea deal with the prosecutor even if you were clearly in the right.
I would say if the guy lunged at you with a knife and you drew your gun and he ran away, you'll be fine if there are witnesses.
I'm familiar with the idea of "brandishing" being illegal, but I looked it up anyway. I found https://en.wikipedia.org/wiki/Menacing which says "Self-defense is often explicitly given as an exception."
It seems like then it would be legal to draw a gun without firing if it was in self defense? I have a hard time believing that there are any cases where shooting someone in self defense would be legal, but scaring them away wouldn't.
This is not exactly what you’re referencing but I bring it up to show just how complicated things can be: Minnesota recently ruled that you do not have the right to use deadly force if you have the opportunity to escape.
And this is the crucial bit, quoting the article: “The court decided the principle also applies to people who merely use the threat of force — meaning one cannot pull a weapon in self-defense if there are other means to escape, even if the person is threatening them with death or bodily harm.”
"There are various situations in which brandishing a firearm may occur. For instance, pulling out a gun during an argument or altercation with another person could be considered brandishing. Additionally, waving or gesturing with a firearm in a manner that suggests aggression or hostility towards others may also constitute brandishing. It is crucial to understand that even if no shots are fired, and no physical harm is inflicted, the mere act of displaying a weapon in such circumstances can lead to serious legal consequences."
I get that LLMs are a black box in ways that most other technologies aren't. It still feels to me like they have to be okay with abstracting out some of the details of how things work.
Unless they have a lot of knowledge in electrical engineering/optics, the average user of this isn't going to understand how the camera or projector work except at a very high level.
I feel like the problem with LLMs here is more that they are not very predictable in their output and can fail in unexpected ways that are hard to resolve. You can rely on the camera to output some bits corresponding to whatever you're pointing it at even if you don't know anything about its internals.
Building projection optics is a bench top experiment that we did in 7th grade in school. Electric circuitry isn't exactly rocket science, either. Things like LCD panels for projecting arbitrary images and CCD chips for cameras become harder to understand.
But the point is to make users understand the system enough to instill the confidence to change things and explore further. This is important because the true power of computer systems comes from their flexibility and malleability.
You can never build that level of confidence with LLMs.
I'm only just now reading about Dynamicland for the first time, so maybe I'm not understanding something obvious. The text description is not very helpful, as far as I can tell from pictures it's a place where you can move around physical objects and papers to do computer programming type stuff?
Under visibility they say:
>To empower people to understand and have full agency over the systems they are involved in, we aim for a computing system that is fully visible and understandable top-to-bottom — as simple, transparent, trustable, and non-magical as possible
But the programming behind the projector-camera system feels like it would be pretty impenetrable to the average person, right? What is so different about AI?
Dynamicland is bootstrapped in a sense, [0] the same way you write the first compiler/interpreter for your code in another language then later write it in it's own language. The code running the camera and projector systems is also running from physically printed programs in one of the videos you can see a wall that's the core 'OS' so to speak of Dynamicland.
I think the vision is neat but hampered by the projector tech and the cost of setting up a version of your own, since it's so physically tied and Bret is (imo stubbornly) dedicated to the concept there's not a community building on this outside the local area that can make it to DL in person. It'd be neat to have a version for VR for example and maybe some day AR becomes ubiquitous enough to make it work anywhere.
[0] Annoyingly it's not open sourced so you can't really build your own version easily or examine it. There have been a few attempts at making similar systems but they haven't lasted as long or been as successful as Bret's Dynamicland.
That's pretty cool. I figure this is explained in some of the videos but I can't watch them right now.
I'm reading more about the "OS" Realtalk
>Some operating system engineers might not call Realtalk an operating system, because it’s currently bootstrapped on a kernel which is not (yet) in Realtalk.
You definitely couldn't fit the code for an LLM on the wall, so that makes sense. But I still have so many questions.
Are they really intending to have a whole kernel written down? How does this work in practice? If you make a change to Realtalk which breaks it, how do you fix it? Do you need a backup version of it running somewhere? You can't boot a computer from paper (unless you're using punch cards or something) so at some level it must exist in a solely digital format, right?
Yeah he's put out a fair number of videos and the whole idea makes more sense there or if you can manage to visit in person.
I think even if you could squeeze down an LLM and get it to run in realtalk I don't think it fits with the radical simplicity model they're going for. LLMs are fundamentally opaque, we have no idea why they output what they do in the end and can only twiddle the prompt knobs as a user which is the complete opposite direction from a project that refuses to provide the tools to build a version because it's putting the program back into the box instead of fileted out into the physical instantiation.
I wish he'd relent and package it up in a way that could be replicated more simply than reimplementing entirely from scratch.
I'm not sure where to draw the line between Realtalk and the underlying operating system. I'm willing to give it some credit, it's interesting without being written entirely from scratch. IIRC most of the logic that defines how things interact IS written in Realtalk and physcially accessible within the conceptual system instead of only through traditional computing.
I believe it runs on stock Linux, but it's an operating system in the sense that it multiplexes tasks and facilitates IPC. The closest analogy I can think of is something like a lisp machine or Smalltalk, where the line between program and OS is really blurry.
Also, if you haven't heard of folk computer[1] as a viable alternative, I'd highly suggest checking it out! I'm one of the contributors, and it's definitely not dead (unlike all the other dynamic land spin-offs I've seen). The head programmers—Omar and Andreas—both worked at dynamic land for a couple months, so they've been able to carry over the good parts while also open sourcing it. The implementations have definitely diverged, but imho in a good way—folk computer is working on multi threading and is much more realtime-safe (you'll see in the latter dynamic land videos that it pauses every second or so).
I'll have to take a look and scrounge up a decent projector to try. The other Bret project I was actually even more interested in was the robotics lab project when I was on a big robotics kick in the 2010s.
As a non-scientist, I am not really any more qualified to give an opinion on this than any random person, but the fear around food dyes seems way overblown IMO.
You can read the statement from the FDA where they banned red dye 3, it's very short. [1] Here's a relevant quote:
>claims that the use of FD&C Red No. 3 in food and in ingested drugs puts people at risk are not supported by the available scientific information.
As a scientist there are a few things to drive my opinion it’s overblown:
- you can’t prove a negative (“this dye isn’t harmful”), all you can do is run a panel of tests and interpret the data
- food additives are tested in animal models at levels that are several orders of magnitude higher than what any human might consume. Animals are then autopsied to see if there are *any abnormalities in any organ system. This is done with several species.
- Cell models are also used to test things like carcinogeniticity, cell-specific toxicity, toxicity of chemicals formed when the additive breaks down, toxicity of trace impurities, etc. It’s quite extensive.
- Data is rarely 100% clear. You may get a signal in some animal model at 1000x expected exposure. What does it mean? Plenty of animals exhibit toxicity not seen in humans and slight abnormalities may or may not translate to humans. But the FDA tends to err on the side of caution, especially with food additives as there is little benefit to offset any risk.
- It’s not unusual to run 10 studies, find 9 are negative, 1 shows a signal but it’s not statistically significant. What RFK tends to do is cherry pick the 1 study and say “there is data to prove it’s harmful!” That’s not how science works. You look at the totality and quality of the data and make the best conclusion you can. Is it 100% foolproof? Of course not, but it’s pretty solid evidence that likely no harm will result.
- The one risk is the “unknown unknowns”. If you don’t know what to look for, you’ll never find it. But that’s true with everything we ingest - drugs, natural foods (peanuts and aflatoxin!), synthetic chemicals, water purification chemicals, etc, etc. We can only do the best with the knowledge we have.
- If you see 10 studies and 5 are positive (barely) and 5 are negative, either the effect is really small (I.e. you should worry more about other things) or it’s just noise.
I have only vaguely heard of this substance so I may be missing some important context, but I don't think this attitude really holds up under scrutiny.
There's plenty of research done on things which can't be patented or used to turn a profit in some way. People do research on diet, exercise, vitamins, and pharmaceuticals which are now generic like aspirin etc. just to name a few off the top of my head.
There's also public funding available for research which isn't intended to make money for any particular corporation.
Damn, I just had one of those moments where you go from thinking you understand something to realizing it's really complicated and you don't understand it at all.
1. It's grown using bacteria, which are much faster to reproduce and cultivate than animal cells. 2. The intended products are a homogeneous mixture, which means you don't need to worry as much about texture