lordswork

fun idea but a bit repetitive and boring.

Technically, yes, but it's a similar relationship of humans being animals. If you say animals, the audience will assume you're not talking about humans.

As long as you're willing to wait up to an hour for your GPU to get scheduled when you do want to use it.

https://www.conservation.ca.gov/cgs/tsunami/maps is getting a hug of death :(

If anyone gets on, please post a screenshot.

Agreed, it sounds like nothing of value was actually lost, just a vibe coded app and synthetic database.

The database was synthetic and vibe coded itself:

he said that the AI made up entire user profiles. "No one in this database of 4,000 people existed," he said.

/compress is the command to do this in most cli agents

Far better than xAI's data center being powered by mobile Diesel generators

This comment got me wondering whether loss of farmland in the US is a serious issue.

It looks like there's about 800 million acres of farmland in the US and we're losing about 2 million acres per year to the land being repurposed. Despite that, crop production has more than tripled in the past 70 years due to technological advances.

That said, economic effects, loss of farmland, and climate change have contributed to slower growth and higher variability of crop yields recently.

In the past decade there's been a modest 0.8% annual increase in crop production despite losing about 2 million acres per year.

For those wondering, the Miyawaki method differs from normal planting by doing the following:

- heavily pre-treat soil with organic matter (simulating forest floor)

- plant a mix of native plants that will make up canopy, tree, sub-tree, and shrub layers

- densely planting plants (3-5 saplings / m^2)

- heavy mulching after planting (weed suppression, moisture control, nutrients)

This encourages rapid growth into a biodiverse dense forest much faster than standard planting techniques.

MOE as an idea specific to neural networks has been around since 1991[1] . OP is probably aware, but adding for others following along, while MoE has roots in ensembling, there are some important differences: Traditional ensembles run all models in parallel and combine their outputs, whereas MoE uses a gating mechanism to activate only a subset of experts per input. This enables efficient scaling via conditional computation and expert specialization, rather than redundancy.

[1] https://ieeexplore.ieee.org/document/6797059

Emissions are trending downward because of shift from coal to natural gas, growth in renewable energy, energy efficiencies, among other things. Major oil and gas companies in the US like Chevron and ExxonMobil have spent millions on lobbying efforts to resist stricter climate regulations and fight against the changes that led to this trend, so I'd say they are the closest to these evil corporations OP described. Additionally, the current administration refers to doing anything about climate change a "climate religion", so this downward trend will likely slow.

The climate regulations are still quite weak. Without a proper carbon tax, a US company can externalize the costs of carbon emissions and get rich by maximizing their own emissions.

I'm most interested in building cool projects, and I have found AI to be a major multiplier to that effort. One of those cool projects was a custom coding tool, which I now use with all my projects, and continue to polish as I use it.

As far as productivity, it's hard for me to quantify, but most of these projects would not be feasible for me to pursue with my limited free time without the force multiplier of AI.

Would you mind sharing what the project is

rust + wasm simulation of organisms in an ecosystem, with evolving neural networks and genes. super fun to build and watch.

which AI you are using?

using chatgpt/claude/gemini with a custom tool i built similar to aider / claude code, except it's very interactive, like chatting with the AI as it suggests changes that I approve/decline.

No sign so far of AI's usefulness slowing down as the complexity increases?

The AI is not perfect, there are some cases where it is unable so solve a challenging issue and i must help it solve the issue. this usually happens for big sweeping changes that touch all over the codebase. It introduces bugs, but it can also debug them easily, especially with the increased compile-time checking in rust. runtime bugs are harder, because i have to tell the ai the behavior i observe. iterating on UI design is clumsy and it's often faster for me to just iterate by making changes myself instead.

I'm at 3k LOC on a current Rust project I'm mostly vibe coding with my very limited free time. Will share when I hit 10k :)

This kind of error doesn't really matter in programming where the output can be verified with a feedback loop.

Did Mirascope exist before MCP? MCP was announced in November, 2024.

Some statistics on US health and safety before and after OSHA:

https://www.perplexity.ai/search/can-you-give-me-an-apples-t...

This has been my bible for performance work internally at Google. Kind of surprised they released it publicly, but I guess they removed all the Gemini-specific details.

I'm building this for myself because I want it too. I recommend you do the same, because it's been really fun to build and teaches you a lot about what LLMs can and cannot do well.

I plan to share it on Github, but waiting for my employer's internal review process to let me open source it as my own project, since they can legally claim all IP developed by me while employed there.

The one that stands out most to me is that it doesn't bundle the AI features with the editor. This is a standalone application that runs on the side of your editor/tools of choice.

This looks very useful.

Unimportantly, I wonder how the copyright date is 2023 when MCP didn't exist until 2024.

well, what are the ideas?

The future is awesome. Thanks for building this!

Why is there no pictures of the actual headset anywhere?

Does recur depend on node?

Not to take away from the main story, but the student was clearly cheating on their homework with Gemini, directly pasting all of the questions.

I like that better too, but open source is often used synonymously with open architecture, so it's good to be aware of.

When it comes to how "open" LLMs are, there are basically three categories:

- Open Source: The architecture of the model is available, so you can bring your own data and compute to train a similar model.

- Open Weights: The trained model itself is available for you to use. You can download it and run it on your own hardware without needing to train anything from scratch.

- Open Data: You get access to the data that was used to train the model. With this, you can completely reproduce the training process, which is super helpful if you want to fine-tune the model or see exactly how it was trained.

There are some other nuances around hyperparams, training methodology, etc. but that's basically the 3 main categories.

Emissions are the bigger issue. Unfortunately, some sort of mining is still required for emission and emission-free sources of energy.