tjchear

I don’t think it’ll necessarily pop. The excess supply could even induce additional demands in not just LLM or AI but in other verticals. The margins of hyperscalers and infra providers might thin out, but with all that surplus in hardware, they can find new uses. For example, developers rent their own GPU node for cheap capable of running frontier open weight models instead of running local models. Computers and consoles will be cheaper again and more people will want to buy them. Non-LLM related simulations could benefit from excess hardware.

Presumably a common argument for sending a trained human expert in place of a robot is because a human can exercise much better judgment on what to explore and dive deep into on site, whereas doing so via a rover is subject to high latency and low bandwidth. It’d be really cool if LLM (or any AI for that matter) reaches the level of sophistication of an in the field scientist and we can send them instead of humans for 80% of the results at 10% of the cost.

I remember reading a research about using a projector like headlights with a high resolution camera that can capture the position of each raindrop and selectively turn off projection on each raindrop’s position in real time so you can still see clearly at night without being blinded by the reflection from the raindrops. It’ll be cool if they can incorporate that once this headlight projection tech becomes widely available.

Here’s my experience: just yesterday I had to tackle this task that’d have required a backend engineer and a frontend engineer several days, so I tasked several Claude code agents to work on them autonomously. With the time freed up, I didn’t just twiddle my thumbs. I used it to read up on this topic that was making the rounds yesterday and gained a better understanding of it - something hard to do when you juggle both a job and raising a family. I could then reinvest the time I used to learn something by using them in some other projects.

Just my two cents. No matter whether you use AI or not, I’m sure you’ll gain something.

Yay thank you!

Lots of fun questions! Can you make it so that I can open each one in a new tab? Also if I navigate back to the main view I lose my scroll position.

If you’re using it with a local model then you need a lot of GPU memory to load up the model. Unified memory is great here since you can basically use almost all the RAM to load the model.

Yeah it’s all trade offs. If it means I get to where I want to be faster, even if it’s imperfect, so be it.

Humans aren’t without flaws; prior to coding assistants, I’ve lost count of the times my PM telling me to rush things at the expense of engineering rigor. We validate or falsify the need for a feature sooner and move on to other things. Sometimes it works sometimes a bug blows up in our faces, but things still chug along.

This point will become increasingly moot as AI gets better at generating good code, and faster, too.

I take a fairly optimistic view to the adoption of AI assistants in our line of work. We begin to work and reason at a higher level and let the agents worry about the lower level details. Know where else this happens? Any human organization that existed, exists, and will exist. Hierarchies form because no one person can do everything and hold all the details in their mind, especially as the complexity of what they intend to accomplish goes up.

One can continue to perfect and exercise their craft the old school way, and that’s totally fine, but don’t count on that to put food on the table. Some genius probably can, but I certainly am not one.

I’ve not used duckdb before nor do I do much data analysis so I am curious about this one aspect of processing medium sized json/csv with it: the data are not indexed, so any non-trivial query would require a full scan. Is duckdb so fast that this is never really a problem for most folks?

My understanding is food delivery companies take a huge cut (like 30%) so restaurants are forced to raise their prices significantly or risk losing customers. Even with that cut, food delivery customers still have to pay a significant delivery/service fee.

iirc there was a startup doing the same thing called sourcetable? How does pane compare to it?

The author makes a good point about language capabilities enabling certain libraries to be written, just as DSL makes it easier to reason about problems and implement solutions with the right kind of abstractions and language ergonomics (usually at the expense of expressivity and flexibility).

There’s a time in my life where I designed languages and wrote compilers. One type of language I’ve always thought about that could be made more approachable to non technical users is an outline-liked language with English like syntaxes and being a DSL, the shape of the outline would be very much fixed and on a guardrail, and can’t express arbitrary instructions like normal programming languages, but an escape hatch (to more expressive language) for advanced users can be provided. An area where this DSL can be used would be common portal admin app generation and workflow automation.

That said, with the advent of AI assistants, I’m not sure if there is still room for my DSL idea.

I think this question can be answered in so many ways - first of all, piling abstraction doesn’t automatically imply bloating - with proper compile time optimizations you can achieve zero cost abstractions, e.g C++ compilers.

Secondly, bloated comes in so many forms and they all have different reasons. Did you mean bloated as in huge dependency installs like those node modules? Or did you mean an electron app where a browser is bundled? Or perhaps you mean the insane number of FactoryFactoryFactoryBuilder classes that Java programmers have to bear with because of misguided overarchitecting? The 7 layer of network protocols - is that bloating?

These are human decisions - trade-offs between delivering values fast and performance. Foundational layers are usually built with care, and the right abstractions help with correctness and performance. At the app layers, requirements change more quickly and people are more accepting of performance hits, so they pick tech stacks that you would describe as bloated for faster iteration and delivery of value.

So even if I used abstraction as an analogy, I don’t think that automatically implies AI assisted coding will lead to more bloat. If anything it can help guide people to proper engineering principles and fit the code to the task at hand instead of overarchitecting. It’s still early days and we need to learn to work well with it so it can give us what we want.

If we sees ourselves less as a programmer and more as a software builder, then it doesn’t really matter if our programming skills atrophy in the process of adopting this tool, because it affords us to build at a higher abstraction level), kind of like how a PM does it. This up-leveling in abstractions have happened over and over in software engineering as our tooling improves over time. I’m sure some excellent software engineers here couldn’t write in assembly code to save their lives, but are wildly productive and respected for what they do - building excellent software.

That said, as long as there’s the potential for AI to hallucinate, we’ll always need to be vigilant - for that reason I would want to keep my programming skills sharp.

AI assisted software building by day, artisanal coder by night perhaps.