Artificial intelligence is good at doing boring repetitive tasks or point out things for a human to look at when there is too much data for us to scrutinise.

If you do not have the skill to verify the output you are almost certainly getting garbage out of it.

The fact that they cram AI into everything is because they are throwing AI at the wall and seeing what sticks. Spending millions on a 1% chance that you will make billions makes economical sense if you can afford to lose.

It’s like a calculator.

• it costs more than doing it yourself
• it seems faster
• it’s accurate but you have to do it right and check the shit after
• it makes you stupid for the basics
• it robs kids of important fundamentals.
• it’s pervasive. Soon it may be ubiquitous.

My niece can’t make change for a dollar, nor multiply numbers above 12. She struggles with writing and reading, and I wonder at her reasoning skills. She’s 27.

She’s fine with a calculator and hard facts. But the “two trains leave on coursetl to pass each other” kind of “make math from description” part that exercises the brain’s ability to select the right equation and use it? Not great.

We’ll be like this, but with building digital things. Our BEST hope is to become the fat and lazy pampered pets of the robots like in WALL-E. The alternative is a little too Matrix for me.

I’ve been doing scientific software development for 20 years, and here’s my personal experience:

• Software development as it had been is already over due to AI. There’s no reason to write code yourself anymore in 99% of cases. Even highly technical stuff (GPU algorithms, for example) are no problem for the AI.

• Scientific work is very different than it used to be. The AI doesn’t have the initiative to just do it for you yet the way it can with writing software, but it’s an expert in every field so it provides an excellent foundation for your own novel ideas and it can often suggest new ideas from fields you aren’t personally knowledgeable about.

• Socializing is going to be very different soon. Right now the AI is already more interesting to talk to than most humans, but talking to it doesn’t feel like talking to a human and doesn’t satisfy the need for interpersonal interaction. I expect that to change. (I suspect that the current absence of a human-like AI is more because developing it is not good business than because it’s technically difficult.)

• A lot of AI critics have objectively false ideas of how well AI works. I think they either haven’t tried cutting-edge models recently or they have deliberately tried to make those models fail. You can break most things if you try; that doesn’t mean they don’t work well.

As I type this, an AI is working on a tricky problem in molecular dynamics for me. It has already found and fixed issues with the algorithm that my co-worker spent a month on, but it thinks that a different algorithm it came up with can do even better. After coming up with the idea, it’s independently implementing it, unit testing it, running simulations with it, and doing statistical analysis of those simulation results. If it gets errors, it can debug them on its own. If the results aren’t as good as it expected, it can iterate over different variants on its own. If the algorithm ends up not working, it can reason about why, from a scientific perspective. What the AI does in hours would have taken me weeks if I came up with the idea, which I didn’t.

Right now human experts can still do what AI does, just slower. I expect that in less than 5 years, that will probably no longer be true. Predicting the future after that point is nearly impossible. Will we all die? Maybe.

Currently? Frank Herbert

Despite the apparent speed of advances, AI is a technology that is still in its childhood. It’s as if tech companies found a way to create a 6-year-old out of GPUs and way too goddamn much electricity. If you can overlook the resource usage required, that’s kind of cool.

The problem is that those same tech companies (and some governments) have decided that they all need their own synthetically created 6-year-old and that they should give it unlimited funding and the ability to make life-changing decisions for everyone else, and anyone who isn’t jumping at the chance to have their life overseen by a machine-generated child with poor impulse control and a propensity to hallucinate is just “getting in the way of progress”.

I think that LLMs are very useful for specific tasks, but they’re being pushed as if they’re useful for almost everything.

I also have huge ethical concerns over LLM-peddlers’ treatment of the environment.

I’m alarmed by some people’s susceptibility to falling for obvious slop and worried about what will happen if and when the slop gets less obvious.

I’m frustrated with many people who share those concerns conflating LLMs with the general concept of intelligence that is artificial.

We have it at work.

I’m not going to lie, it has made my job a lot easier. My fear is it’ll make me stupid so I do most of my work myself and then get the LLM to tweak it.

As an example, I’ve had to create a number of policies and procedures.

I’ve done the content, the LLM has done all the formatting etc. Something that would have taken me several hours to do. I’d prefer an assistant to do it for me but thats not gonna happen.

You are not going to get a general pulse here. This is a fuck ai echo chamber. I am not saying that the sentiment is wrong, but it isn’t representative of the general public.

Currently, it’s a skill amplifier. You can’t code for shit and rely solely on it, it will give a shit product. You know how the whole codebase works and use it as needed? It will speed you up drastically while maintaining the quality you want.

Agents are a nightmare. They introduce a variable completely uncontrollable by you to your project, which will make you lose track of how everything works, resulting in a shit product.

I see its future on more optimized models that can run well locally on lower hardware, not those that use entire datacenters to power it. The big AI companies still do not generate profit and are doomed to fail in long term, but AI itself isn’t going away.

• How come the most memory-inefficient ML model we have ended up being the most powerful one. Can we rewrite the universe so modern advanced AI is based on network inference or ensemble models or sth
• How come AI has taken over the economy, and it doesn’t seem like a bubble (or at least not the crypto type of bubble)
• I’m a bit surprised that the most predominant use of AI/LLM seem to be… as a search engine wrapper. Thankfully ChatGPT is getting better at it, their default model has a built-in search engine integration now. This is kind of important since the company is getting big enough that their model bears social responsibilities…
• I have an asshole-ish work environment where the head honchos only cares about professionalism and efficiency… so I have no shame using AI for work. They don’t deserve my best tbh
• In terms of actual usage for my work, AI outputs can be anywhere from very good to very bad in an instant, and sometimes it’s hard to tell the two apart…
• It’s surprisingly helpful as a primer to help me with things I’m not good at; researching niche information, finding resources
• For personal usage, I’m still hoping that there is a way to work AI into some type of a home automation loop… I haven’t found that sweet spot yet. Maybe it’s just my lack of skill at the moment. Hopefully I can hook together some shitty models with search engine skills (on my own SearXNG) as a semi-useful agent

I refuse to use AI as a tool for anything other than making tools.

I can’t write LISP but I use AutoCAD. I use an LLM to write LISP routines.

I spent MONTHS looking for a way to catalog my plant collection. Tried a dozen open source projects and none did what I wanted it to do. So I used Claude to make a simple database website…and it was able to add a ton of pipe dream features.

Nothing I make is for sale and I don’t claim it as my own. They’re tools for ME to use. And the important part is that once the tool is done, the LLM gets closed.

It works for basic tasks like correcting grammar, sorting lists, etc, but it makes too many mistakes when you try to use it for a complex task.

I feel like it’s making people dumber. I keep getting screenshots of Google ai summary from people as if that’s the answer and they’re confident in it. I keep finding those are not correct. For me, ai has created more work because I can no longer trust my colleagues statements.

My opinion is well… I’m sure it may have some use, but I see it pushed in the dumbest and least useful ways. It is fundamentally incapable of creating art in any shape or form. Use it to summarise things? I don’t trust it enough to correctly summarise the text. Use it to write/reply to emails? I’d myself just feel insulted if someone sent me a generated email. Why should I expect someone to read a text I didn’t even bother to write?

The small help it gives does not equal the vast evil done to the environment, communities, and actual makers and creators.