mattbrewsbytes

I read somewhere that a professor is taking the approach with some assignments to ask students to have AI generate their paper, then ask for references and then evaluate the AI output. They were finding this is a good way for students to learn the material because they have to understand it well enough to scrutinize the generated text.

I think this is similar in concept to after you've learned something to teach or explain it to someone else and it helps you solidify your knowledge.

Generative AI is a tool, much like when search engines and massive internet use came about in the late 90's I'm sure college kids were sharing completed papers online and searching for them. Learning how to effectively use the tool in your daily work or learning is key, relying on the tool to do your work blindly, which is likely what the CA colleges are trying to prevent, isn't going to help learn the material.

Sounds like their plans are to make it like an exhibit and/or field trip for kids to learn about aviation.

It would be neat if they refurbished the cockpit to make it operational and changed the windscreens to computer monitors, basically make it a low-feature simulator that people could experience.

Can switches like this also detect usage, like current flowing? It might be a fun and healthy twist to plug a walking pad/treadmill into a smart plug and then either make it so brainrot websites are only viewable when the treadmill is being used or make a small app that tracks the time the treadmill is used and then only allows that amount of time on brainrot websites.

Selling point from a developer and career perspective for sure. Its also fun to program in and at least for me made me think about solutions differently.

Its a negative point for engineering leaders that are the decision makers on tech stacks as it relates to staffing needs. LLMs not writing it well, developers that know it typically needing higher compensation, a DIY approach to libraries when there aren't any or they were abandoned and haven't kept pace with deprecations/changes, etc.

In the problem space of needing a web framework to build a SaaS, to an engineering leader there are a lot of other better choices on tech stack that work organizationally better (i.e. not comparing tech itself or benchmarks, comparing staffing, ecosystem, etc.) to solve web SaaS business problems.

I don't know where I stand personally since I'm not at the decision maker level, just thought I'd point out the non-programmer thought process I've heard.

Buy a minimal set of tools and start flipping cars. Then try to get a dealership license in your state. Start with small fixups and detailing work, flip one car, use the proceeds to get something a little more valuable, repeat.

Be honest with people when selling, no slimy sales tactics, ask for referrals. Maybe every few cars find someone who is in desperate need and sell it at cost, ask for referrals. Buy some more tools, repeat. Maybe get a small shop with a lift. Maybe hire a certified tech for really tough problems, someone who values integrity and doing a job right, like a semi-retired person that knows a lot.

Grow that $35k into a viable, honest, local, AI-proof business.

I'm not expecting perfect output, if there is such a thing. I'm expecting output equivalent to a junior engineer or intern at this point in time. I can tell an agent in the editor I want it to add a many to many join relationship and a multi select field to a form and it does it pretty decently. In some ways this is more automating the tedious tasks which is really what AI's strong suit is right now, in my opinion. Sure I could go move a method around, refactor the 5 classes that call it, etc. but I can delegate that task and its done in a few seconds while I think about the next thing I want to do.

I find if you get too lengthy with tasks or too advanced AI goes off the rails so to speak, sorry for the pun.

With so many generative AI tools out there, picking more obscure or niche languages is a detractor because the AI models won't have as much depth of training to pull from when you ask it to do things.

Also, technology choices for B2B web apps is rarely going to be a sole factor in determining success or failure of any business. As much as this community likes to compare performance metrics, benchmarks, frameworks and everyone has personal tastes on what is "good", all of those discussions are mostly irrelevant. Picking something the team is comfortable with and has depth of knowledge in is a good practice.

So just pick Rails and move on with solving business problems :)

That is entirely where I'm coming from - plane designs from the past had longer wingspans and supported 4 engines. Engines have gotten more efficient and powerful so cross-ocean routes don't need 4 engines anymore.

So I was wondering what if current plane designs had 4 smaller but equally as efficient engines instead of 2. The vast majority of airplane accidents happen at takeoff or landings, if some of those can be avoided by having 4 engines for commercial aircraft, its a worthwhile idea to explore.

Also I would venture that 99% of comments on public forums like this are from people without expertise. My expertise in this space is a curiosity about planes for a few decades, taking some actual flying lessons, and being generally interested in aviation to go to airshows, watch youtube content from pilots, etc. I probably have about the same aviation knowledge as an average HN person.

Apple is absolutely into the mature company phase if what they are coming up with is essentially a different bevel on buttons that "makes apps and system experiences more expressive and delightful". They are the IBM of the 90's at this point, way overvalued stock, not much room there for growth.

Setting that aside, video game visuals serve a different purpose than application UI widgets for menus, forms, buttons, etc. In a video game there are different reasons to highlight objects in a game world and relative to other objects and their movement or ability to move/interact with them. Objects in a game world might have their own light source, move in different ways so that the player knows its something they can interact with. If you look at open world games there are a lot of things happening visually in varying degrees of grabbing the gamers attention.

In a UI that is task oriented, like operating systems for PCs or phones, there needs to be less visual clutter, its not a game. It should be obvious to massive numbers of the population what function a UI element serves and accessibility standards come into play because often these OSes are used in corporate environments which should allow for people with different visual abilities to do their work.

I was just watching something the other day about how jet engines have gotten more efficient and powerful over the last 50 years where commercial airliners really only need 2 engines. All 2 engine aircraft also have to be able to operate on 1 engine as well if there is a failure.

One has to wonder if this was a bird strike incident on both engines that maybe having 4 engines would have allowed the plane to circle back around.

Revolutionary technology I think needs to be something that gets across that early adopter chasm. We haven't really seen any 3D or goggle tech do that. There are VR things but I wouldn't consider VR video games a mainstream set of people. Mainstream to me is smartphones, AI usage, etc. things that catch on with the majority of the population.

I think that anything that is going to require humans to wear something on their head for entertainment purposes is not going to make it to mainstream. There will be niche uses and likely video games are a good niche. What is going to compel people to buy something like the Vision Pro when they already have smart phones that can do everything anyways?

Also, humans over 40 tend to start needing reading glasses and with each year of age more and more people need them. Its hard to have any device that covers the eyes also take into account people's vision issues, minor or major.

JIRA gets a lot of deserved hate but I think nearly all software work tracking systems suffer from the same issue: marketing over promises about velocity and predictions about project delivery timelines.

Have we as an industry gotten any better at delivering projects on time? If you have a lot of dysfunction in your organization no software is going to fix that. Or to put it another way, you can’t solve people problems with software.

> This was the craftsman’s ‘Golden Age’ and much time and trouble was taken over the design of tools. Craftsmen were being called upon to do more skilful and exacting work and the use of tools and the interest in development had become very widespread.

Above pulled from A Brief History of the Woodworking Plane [0]. A woodworking tool that has evolved over 2,000 years. Now there are electric planers, handheld electric planers and lots of heavy machinery that do the same thing in a very automated way. If a company is mass producing kitchen cabinets, they aren't hand planing edges on boards, a machine is doing all that work.

I feel like with AI we are on the cusp of moving beyond a "Golden age" and into an "industrial age" for coding, where it will become more important to have code that AI understands vs. something that is carefully crafted. Simple business pressure will demand it (whether we like it or not).

^ A comment I made just yesterday on a different thread.

For software developers AI is like the cabinet maker that gets a machine to properly mill and produce cabinet panels, sure you can use a hand plane to do that but you're producing a very different product and likely one that not many people will care about, possibly not even your employer when they see all the other wood shops pumping out cabinetry and taking their market share.

[0] https://www.handplane.com/879/a-brief-history-of-the-woodwor...

People love their craft and want to really understand how a tool works. They might start with taking a tool apart, looking at how it works, putting it back together and then using it for a bit. Then they need a similar tool for a slightly different use case, so they start making one themselves. Maybe they have more specific, detailed needs than existing tools. Maybe there is something about the tool that annoys them or they think they could make it work better. This keeps going over time.

> This was the craftsman’s ‘Golden Age’ and much time and trouble was taken over the design of tools. Craftsmen were being called upon to do more skilful and exacting work and the use of tools and the interest in development had become very widespread.

Above pulled from A Brief History of the Woodworking Plane [0]. A woodworking tool that has evolved over 2,000 years. Now there are electric planers, handheld electric planers and lots of heavy machinery that do the same thing in a very automated way. If a company is mass producing kitchen cabinets, they aren't hand planing edges on boards, a machine is doing all that work.

I feel like with AI we are on the cusp of moving beyond a "Golden age" and into an "industrial age" for coding, where it will become more important to have code that AI understands vs. something that is carefully crafted. Simple business pressure will demand it (whether we like it or not). The difference with hand planes (and other woodworking tools/concepts) is that a lot of people can appreciate a piece of furniture and the craft that went into it by opening a drawer and seeing hand cut dovetails for example. When it comes to digital products, none of the consumers of the product will really care about the craftsman that created a JS library that they are indirectly using on a web site that has its "view source" hidden with minification/obfuscation.

[0] https://www.handplane.com/879/a-brief-history-of-the-woodwor...

If you're going to analyze a process you have to look at all of the steps involved, including waiting time, re-work or going backwards in the process/iterating, as well as the time to do the tasks. They all matter in the economics of delivery of working software.

The slowest part is going to be the parts involving humans communicating things to each other and iterating on the requirements and designs and the waiting times in between. The technical coding or generation of assets does not take that much time in the overall process.

This is also the case in the past with companies like Xerox and why didn't they do X with Y that came out of PARC.

There's only so much product bandwidth a company can take on that makes sense, look at the graveyard of Google products.

> "Enshittification" is a term coined by Cory Doctorow to describe the gradual decline in quality and user experience of a product or service, particularly online platforms, as they prioritize profits over user satisfaction

- Google Gemini results

How long until AI results/output get the enshittification treatment?

If a tool lets someone turn it into draconian screens filled with fields upon fields and arcane workflows then the hate is warranted.

I have yet to see any organizational improvements in our delivery cadence or value derived from any of the planning elements of Jira. With all the doo-dads and buttons and fields and multiple ways of organizing things Jira is just a PM dopamine fidget spinner.

Post-Its on a whiteboard with dry-erase drawn columns is really all that's needed. Use a wiki for tracking longer lived todo lists and documentation.

I loathe Jira. People should be allowed to have that many fields on a screen, its ridiculous what is allowed. What sets linear apart from it?

Interacting with LLMs or AI APIs sounds like other software patterns, it doesn't matter that its AI or an LLM really, you are calling a function and providing inputs and expecting output. You get better output when your inputs are tuned to the scenario. Some of your inputs in this paradigm could be considered as optional parameters because you still get output without them.

If you need to remember parts of the inputs in between user sessions then you need to save state of those somewhere to a disk. Databases are a common choice especially in web development but you could also just put things in a file. Another option if this isn't a web development context is to use something like sqlite since it will help organize the data a little better than say CSVs or similar.

I think this is the key point - LLMs will go through a commoditization phase and I think you left out a key example from a technology and business context: search engines. There was a huge trend where everyone needed search and was building search, etc. and a couple decades later there are lots of companies that evaporated and a few left standing.

There was also a dot com bubble, mostly bursting not because of search but because there were a lot of what today would be "AI startup" but is just a web app calling AI Api's. So there's likely to be some bubble burst but it should be smaller maybe hitting more of these small tools that eventually become features.

There was a time back in the years after dot com era where search servers were sold with the proprietary software loaded for enterprises to use. A google box running on your local network. Now there are advanced things like elastic and other search specific technologies that you can use.

I think LLMs might follow this market pattern where you can buy something to host yourself and then commoditization happens enough where open source solutions will also evolve to have good enough solutions.

An idea for a disrupting company would be to open source their LLM and offer support and feature development to enterprises as the paid offering, kinda like Red Hat or others doing that model. A key difference is running an LLM locally on decent sized compute is fine but it will be costly to scale on your own.

I know you kid, or at least that gave me a chuckle, but this might be why seniors don’t find a lot of usefulness with AI, because we see blatant issues immediately and think it’s a waste of time sometimes.

This aligns with my thinking that juniors can leverage AI to become seniors much faster than people without AI but there are important concepts about learning debugging, learning how the underlying technologies work, etc. that is crucial to becoming a senior dev. Just using AI to pump out code without knowing the details of how it works will not end up going well for the long haul.

Hey I'm curious how the consumers of those queues typically consume their data, is it some job that is polling, another piece of tech that helps scale up for bursts of queue traffic, etc. We're using the google equivalent and I'm finding that there are a lot of compute resources being used on both the publisher and subscriber sides. The use cases here I'm talking about are mostly just systems trying to stay in sync with some data where the source system is the source of record and consumers are using it for read-only purposes of some kind.

Google PubSub is what we use as our message queue, mostly for communicating change data capture via messages to other internal systems. Its typically being consumed by some job system polling on an interval and then doing CRUD to sync changes.

Its not very complex and feels like we're running a lot of compute resources to just sync data between systems. Admittedly there isn't good separation of concerns so there is overlap that requires data syncs.

I've been looking at things like kafka, etc. thinking there might be some magic there that makes us use less compute or makes data syncs a little easier to deal with but wonder what scale of data throughput is a tipping point where a service like that is really needed. If it turns out its just a different service but same timeliness of data sync and similar compute resources I struggle with what benefits might be provided.

I'd love for almost like a levels.fyi style site where people could anonymously report things like this for the tech stacks being used, throughput of data, amount of compute in play, and ratings/comments on their overall solution ("would do again", "don't recommend", "overkill", "resume filler"). It feels much like other areas of technology where a use case comes out of a huge company and RDD (resume driven development) takes hold and now there are people out there doing the equivalent of souping up a 1997 honda accord like its a racecar but its only driving grandma to her appointments.

This should be a ShowHN, I don't see any actual questions being asked.

Wow, so that RubyLLM gem makes writing an agent more about basic IO operations. I have somehow thought there needed to be deep understanding of LLMs and/or AI APIs to build things like this where I would need to research and read a lot of docs, stay up to date on the endless updates the various AI systems have, etc. The example from the article is about files and directories, this same concept could apply to any text inputs, like data out of a Rails app.

Smart juniors shouldn't be juniors long. I think another angle to think about this is Junior devs that can leverage AI to learn have a much shorter path to Senior than current Seniors had to take (aka the hard way). If as a Junior you know of things to look out for but don't know the implementation details, AI can help you.

For really obscure issues or bugs, I had to scroll through seemingly random StackOverflow results (or Experts Exchange, ha) looking for exact issues that I had, weeding through answers that weren't relevant, things based on older versions, etc. Now you give the context to AI tech stack, data, and your source code, explain the problem with the error message, and ask what might be happening and that shaves a few hours off the old way of trial and error. Even if the 1st answer isn't totally right, a few follow on questions will help narrow it down.