How is this different from running Paseo [0] in your VM?
908 karma · joined March 13, 2024
I think it's the opposite -- if you have a good way to design your software (e.g., conceptual and modular), LLM will generate the understanding as well. Design does not only mean code architecture, it also means how you express the concepts in it to a user. If software isn't really understood by humans, I doubt LLMs will be able to generate working code for it anyway, so we get a design problem to solve.
> Time-boxing AI sessions.
Unless you are a full-time vibe coder, you already wouldn't be using AI all the time. But time boxing it feels artificial, if it's able to make good and real progress (not unmaintainable slop).
> Separating AI time from thinking time.
My usage of AI involves doing a lot of thinking, either collaboratively within a chat, or by myself while it's doing some agentic loop.
> Accepting 70% from AI.
This is a confusing statement. 70% what? What does 70% usable even mean? If it means around 70% of features work and other 30% is broken, perhaps AI shouldn't be used for those 30% in the first place.
> Being strategic about the hype cycle.
Hype cycles have always been a thing. It's good for mind in general to avoid them.
> Logging where AI helps and where it doesn't.
I do most of this logging in my agent md files instead of a separate log. Also after a bit my memory picks it up really quickly what AI can do and what it can't. I assume this is a natural process for many fellow engineers.
> Not reviewing everything AI produces.
If you are shipping in an insane speed, this is just an expected outcome, not an advice you can follow.
(just joking, your posts are great, Simon!)
> The Hacker News front page alone is enough to give you whiplash. One day it's "Show HN: Autonomous Research Swarm" and the next it's "Ask HN: How will AI swarms coordinate?" Nobody knows. Everyone's building anyway.
These posts got less than 5 upvotes, they didn't make it to home page. And while overall quality of Show HN might have dropped, HN homepage is still quite sane.
The topic is also not something "nobody talks about," it's being discussed even before agentic tools became available: https://hn.algolia.com/?q=AI+fatigue
String theory usually prefers universes that want to crunch inwards (Anti-de Sitter space). Our universe, however, is accelerating outwards (Dark Energy).
To fix this, the authors are essentially creating a force balance. They have magnetic flux pushing the universe's extra dimensions outward (like inflating a tire), and they use the Casimir effect (quantum vacuum pressure) to pull them back inward.
When you balance those two opposing pressures, you get a stable system with a tiny bit of leftover energy. That "leftover" is the Dark Energy we observe.
You start with 11 dimensions (M-theory) and roll up 6 of them to get this 5D model. It sounds abstract, but for my engineer brain, it's helpful to think of that extra 5th dimension not as a "place" you can visit, but as a hidden control loop. The forces fighting it out inside that 5th dimension are what generate the energy potential we perceive as Dark Energy in our 4D world. The authors stop at 5D here, but getting that control loop stable is the hardest part
The big observatiom here is that this balance isn't static -- it suggests Dark Energy gets weaker over time ("quintessence"). If the recent DESI data holds up, this specific string theory solution might actually fit the observational curve better than the standard model.
[0] https://ocw.mit.edu/courses/8-821-string-theory-and-holograp...
Low quality engineering is always visible to outside. Low quality engineers using LLMs won't get any better.
I'm not sure what you're referring to. I didn't say anything about capabilities of people. If anything, I defend people :-)
> And yes guard rails can be added easily.
Do you mean models can be prevented to do dumb things? I'm not too sure about that, unless a strict software architecture is engineered by humans where LLMs simply write code and implement features. Not everything is web development where we can simply lock filesystems and prod database changes. Software is very complex across the industry.
Not that humans can't make these mistakes (in fact, I have nuked my home directory myself before), but I don't think it's a specific problem some guardrails can solve currently. I'm looking for innovations (either model-wise or engineering-wise) that'd do better than letting an agent run code until a goal is seemingly achieved.
Day 1: Fed. (Inductive confidence rises)
Day 100: Fed. (Inductive confidence is near 100%)
Day 250: The farmer comes at 9 AM... and cuts its throat. Happy thanksgiving.
The Turkey was an LLM. It predicted the future based entirely on the distribution of the past. It had no "understanding" of the purpose of the farmer.
This is why Meyer's "American/Inductive" view is dangerous for critical software. An LLM coding agent is the Inductive Turkey example. It writes perfect code for 1000 days because the tasks match the training data. On Day 1001, you ask for something slightly out of distribution, and it confidently deletes your production database because it added a piece of code that cleans your tables.
Humans are inductive machines, for the most part, too. The difference is that, fortunately, fine-tuning them is extremely easy.
FYI, back raises have been really helpful.
The value of a Bloomberg Terminal isn't the UI (which is famously terrible/efficient); it's the latency, the hardware, the proprietary data feeds, the chat network, and the reliability.
Building a React frontend that fetches some JSON from an API in 2 hours is impressive, sure, but it’s not the hard part of fintech. We need to stop conflating "I built a UI that looks like X" with "I rebuilt the business value of X."
This is the part that terrifies me. Generating code has never been the bottleneck; understanding it has. If you aren't reviewing the code, you are effectively introducing a black box into your stack that you are responsible for but do not understand.
While on point, I like the patterns Effect is introducing, but I have already been using them pre- and post-AI for a long time now (especially in TypeScript). It's not an innovation. We shouldn't be surprised when LLMs are better at working on robust code bases with modular design.
Separately, be cautious of people putting illegal content on your platform.