One solution: do NOT just program for work. If it's not work related - where management can dictate how you work - you can whatever you want, and if what you want is to keep writing software and not outsource your brain to an AI, absolutely do so.

I’ve come to the same conclusion, though my line of work was research rather than software engineering. “He who pays the piper calls the tune.” It’s fun as long as I enjoyed the tunes being called, but the tunes changed, and I became less interested in playing.

I am now a tenure-track community college professor. I’m evaluated entirely by my teaching and service. While teaching a full course load is intense, and while my salary is nowhere near what a FAANG engineer makes, I get three months of summer break and one month of winter break every year to rejuvenate and to work on personal projects, with nobody telling me what research projects to work on, how frequently I should publish, and how fast I ship code.

This quote from J. J. Thomson resonates with me, and it’s more than 100 years old:

"Granting the importance of this pioneering research, how can it best be promoted? The method of direct endowment will not work, for if you pay a man a salary for doing research, he and you will want to have something to point to at the end of the year to show that the money has not been wasted. In promising work of the highest class, however, results do not come in this regular fashion, in fact years may pass without any tangible results being obtained, and the position of the paid worker would be very embarrassing and he would naturally take to work on a lower, or at any rate a different plane where he could be sure of getting year by year tangible results which would justify his salary. The position is this: You want this kind of research, but, if you pay a man to do it, it will drive him to research of a different kind. The only thing to do is to pay him for doing something else and give him enough leisure to do research for the love of it." (from https://archive.org/details/b29932208/page/198/mode/2up).

That was the original strategy for universities: teaching was the job, and research was the side-product of having some very smart people with free time. Until some "genius" decided that it was better to have professors competing for money to pay directly for their research. This transformed a noble and desirable profession into just another money searching activity.

Fine that doesn't change the fact for a lot of people they felt they had "if you love what you do you don't work a day in your life" and now they don't. They aren't wrong to feel a sense of deep loss.

I wholeheartedly agree. Computing professions such as software engineering used to feel like, "Wow, they're paying me to do this!" Yes, there was real work involved, but for many of us it never felt like drudgery, and we produced, shipped, and made our customers, managers, and other stakeholders happy. I remember a time (roughly 20 years ago) when zealous enthusiasts would proudly profess that they'd work for companies like Apple or Google for free if they could work on their dream projects.

Times have changed. The field has become much more serious about making money; fantasies about volunteering at Apple have been replaced with fantasies about very large salaries and RSU grants. Simultaneously (and I don't think coincidentally), the field has become less fun. I recognized how privileged this sounds talking about "fun", given how for most of humanity, work isn't about having fun and personal fulfillment, but about making the money required to house, feed, and clothe themselves and their loved ones. Even with the drudgery of corporate life, it beats the work conditions and the abuse that many other occupations get.

Still, let's pour one out for a time when the interests and passions of computing enthusiasts did line up with the interests of the corporate world.

The money was what did it, not the AI. If we were all just tinkering with the AI all day long this stuff would still be fun.

Money sucking the joy out of things, a tale as old as time.

I didn't say they were. I feel it too.

Or just shut down your computer after work and “touch grass”. Go to the gym, hang out with friends and family.

My “brain” has always been a systems thinker. I was fortunate enough even in my first job to be directly in front of our customer and gathering requirements, not having the label for it then but trying to solve XYProblems, dealing directly with users and their pain points and seeing an entire data entry department built around my code. This was when I was 22 - 3 decades ago.

Now my brain helps me go from ambiguous, conflicting requirements, working with people, an empty AWS account and an empty git repo to a complete working solution.

Coding has always been the necesary grind between vision and implementation

Totally agree. I'm baffled by those who don't clearly see that Codex works better than C.C. in many ways.

Codex being faster is not at all equivalent to working better. Claude Code does what I need from it most of the time.

I said better, not faster. Primarily, it writes better code, works more smoothly, is better at describing what it's doing and what it's done.

All lower case, instant won't read.

Agree. It is just stupid, doesn't serve any functional purpose.

In this case, it seems to be serving as a useful filter function.

Yeah, I got the sarcasm, but that's ok.

I'll water my indoor tomatoes, basil and thyme. It is way more productive than blabbering about gardening in the internet.

I didnt even notice.

I have an RTX 6000 Pro Max-Q, which has 96GB VRAM. It identified the hardware correctly but incorrectly thought it had 4GB, at least if I interpret the RAM dropdown correctly.

Then it shows the full resolution models, which are completely unnecessary to run quality inference. Quantized models are routine for local inference and it should realize that.

Needs work.

This is really important work.

Try Meetup.com. It's unfortunately gone downhill in many ways, but there are still people using it. Search for in-person meetups in your area and if any look interesting, sign up and go to some events.

Yep, meetup.com, eventbrite.com and for the musically inclined residentadvisor.com and dice.fm.

The trick for me was to offer to help with the organization of the event, it is more rewarding and makes for more lasting connections.

Curious why the large majority of the clips are about water. If that's spontaneous, maybe it's interesting psychologically.

If you walk very early in the morning looking for something to capture while everyone is asleep, chances are the water bodies (ocean, river, lake, ..) are somewhere you know that you can plan to go to, and also they are reliably in the same place - so if say you forgot and took dead batteries, your cable failed or you got the wrong mic, you can try again on the next day.

Capturing other things may be harder - a bird may not be there, the wind may change from day to day and so on. Water is usually satisfying to hear and a good subject to experiment and explore when you are beginning in field recording.

there's just more of them. the freesound api-based parser searches for tags related to nature•

Just want to point out this comment is highly ironic.

This is all a computer does :P

We need llms to be able to tap that not add the same functionality a layer above and MUCH less efficiently.

> We need llms to be able to tap that not add the same functionality a layer above and MUCH less efficiently.

Agents, tool-integrated reasoning, even chain of thought (limited, for some math) can address this.

You're both completely missing the point. It's important that an LLM be able to perform exact arithmetic reliably without a tool call. Of course the underlying hardware does so extremely rapidly, that's not the point.

The computer ALREADY does do math reliably. You are missing the point.

Could you explain why that is?

A tool call is like 100,000,000x slower isnt it?

No idea really, but if it is speed related I would have thought that OP would have used faster rather than importance to try and make their point.

It's both. Being dirrctly a part of it makes it integrated into its intelligence for training and operation.

That would be cool. A way to read cpu assembly bytecode and then think in it.

It's slower than real cpu code obviously but still crazy fast for 'thinking' about it. They wouldn't need to actually simulate an entire program in a never ending hot loop like a real computer. Just a few loops would explain a lot about a process and calculate a lot of precise information.

"This AI child knows about the world more than we do since it has been trained on the whole internet, but it doesn’t have millions of years of evolution, genes, or a nervous system to back up its morality and empathy. This means we need to install morality in AI from scratch. But how do we install something in a software system that we can’t even define ourselves?"

Two things jump out to me. First, there's a vast amount of information about morality in that internet-scale data dump. It's not some mystical thing orthogonal to knowledge, which ties directly into the second thing: maybe the author can't define it, but many others can and do, especially philosophers, though YMMV given which philosophy and philosopher.

viz: "What is morality, or ethics? It is a code of values to guide man's choices and actions—the choices and actions that determine the purpose and the course of his life. Ethics, as a science, deals with discovering and defining such a code."

There are different systems of ethics, and those themselves can be reasoned about. We should want one that values rational human life. And among that vast data dump, there's no doubt that those ethical ideas have been extensively written about - so it can be used as a reference point to actively emphasize and select the ideas.

Have you gotten any nastygrams from Wolfram about this? They're pretty protective of their IP. Not saying I think that it's some violation of it, but I could see them being alarmed.