My Mixed Feelings About AI

I find myself torn when it comes to artificial intelligence. On one hand, I’m genuinely amazed by its possibilities; on the other, I’m deeply concerned about its implications. I’ll admit I’m perhaps more fascinated than worried, which sometimes makes me wonder if I should be more cautious.

Why I’m Concerned

My worries stem largely from the perspective of thinkers like Aurélien Barrau, a French astrophysicist, poet, and environmental activist whose work I deeply respect. He argues that without a fundamental shift in our philosophy of living, our current climate and ecological crisis will only worsen. As long as we prioritize capitalist growth over meaningful human relationships and planetary health, technology—including AI—becomes part of the problem rather than the solution.

Barrau identifies several key concerns that resonate with me:

Geopolitical and military risks: AI could encourage us to delegate critical decisions to machines, especially when speed seems more important than wisdom.

Techno-solutionism: The dangerous belief that technology alone can solve all our problems, when what we really need is to fundamentally rethink our values and way of life.

Social alienation: The potential for AI to isolate us further from genuine human connection and community.

As Barrau puts it: “We’ve built a system where prioritizing life over money appears extreme.” His perspective really captures something profound about our current predicament.

Why I’m Still Fascinated

Despite these concerns, I can’t help but be impressed by AI’s rapid progress. Critics who claimed certain capabilities were impossible have been proven wrong time and again.

I’ve been experimenting with AI as a creative assistant in several areas:

Music: Tools like Suno AI let me generate songs that, while perhaps not professional-quality, are genuinely impressive for someone without formal musical training. I’ll even confess I’ve created a few songs glorifying myself—guilty as charged!

Writing: I’m working on several fantasy and fiction projects that involve extensive worldbuilding—creating detailed magic systems, languages, currencies, characters, and entire civilizations. This process used to take forever, but AI assistance has dramatically accelerated my workflow while still letting me maintain creative control.

Programming: AI excels at repetitive, pattern-based tasks. For instance, it’s revolutionized how I handle database seeding—no more lorem ipsum! I can describe a data model and get realistic test data generated automatically. It’s also made documentation much easier for someone whose native language isn’t English, and I use it to create technical cheat sheets for tools I use occasionally but can’t memorize.

Finding Balance

I don’t want AI to do everything for me—I see myself more as a conductor, guiding and infusing soul into the building blocks that AI helps generate. My hope, perhaps utopian, is that AI can truly serve all of humanity while addressing the ecological concerns that make its development sustainable.

The challenge isn’t the technology itself, but ensuring we develop and deploy it in ways that enhance rather than diminish our humanity and our planet’s health.

I’m genuinely worried about the hype. I was in a conversation yesterday with a few people, some who work for small/medium sized companies and some who work for huge companies and they say the same thing… management comes and says “we want to implement GenAI/LLM/buzzword”. No one knows what that means or what it looks like but no one pushes back. Instead they know they’ll be able to pad their resume with “worked on GenAI” and so they work on dubious business value work and produce worse products/features we all have to end up living with and so continues the cycle of software products getting worse and normalized. And eventually the bubble will pop and people will lose their jobs but of course, not anyone from management.

As far as using it as a developer, I do not have a very positive view of AI. I use the fancy auto-complete to deal with boilerplate but otherwise I never do anything like “Write a function that does X” because thinking through those problems and designing code are my favourite parts of programming. I don’t care that it’s slower and in my view tools that allow less people to ship more bloated software at a faster rate isn’t a good thing. Speeding up your side-project so you can have more time with your family is a much different thing, of course, but that gets into the question of if the time saved will actually be used to do something or just be used to spend more time on the computer (remember when computers were supposed to make us work less?)

If it gets to the point where I can’t get a job because I refuse to “vibe code” or offload everything to an agent then I’ll very likely be leaving the industry.

Caught up with a friend who started grad school after engineering for a decade. Group projects where he was with folks who just came out of undergrad - they didn’t know how to do anything without the AI.

Think every git command, every small tweak, every thing. They were writing a video decoder in python (their choice). It was processing a 300 MB file. Their solution used 8 GB of swap memory.

He spent the entire project time refactoring their slop.

So, I’m pretty worried.

It’s a hot topic, and I tip my hat to some of the nuanced thoughts and replies.

AI can’t be ignored… it’s a powerhouse. The ability to describe a solution and have it magically come up with something that works is amazing. Yes, sometimes the code is slop, but this saves countless hours. The era of highly polished “craft coding” may be over.

However… if we look a bit deeper into this, the tradeoffs that it implies start coming more into focus and it might not be all roses. AI is obliterating many jobs – I won’t waste words listing them, but it’s a lot. Junior developers are getting displaced ferociously, and senior devs may not be that far behind.

What really gets me thinking about this is when I think about my own livelihood as representative for many people whose jobs are being taken over. Where do they (we) go? Are we really expected to pivot careers? That’s challenging in the best circumstances, but with the short timelines that these changes require, it almost becomes an extinction level event for huge swaths of the working class. There may be some jobs that get created as others are destroyed, but I suspect it’s killing more jobs than it is creating… whatever the case, I don’t think you can wave your hands at this. And if huge chunks of the working class become unemployed, then our finances will circle more closely around food and housing budgets. And with higher levels of economic uncertainty, there are higher levels of political instability because people get scared and are more likely to believe the stories from an opportunistic demagogue or authoritarian. AI clearly exacerbates some problematic trends that people have… and whether it offers solutions/improvements to some of the problems facing humanity seems less clear. It seems likely that AI will be a direct or indirect cause of greater instability and economic disparity.

Stay tuned!

Personally I’m trying to stay clear of predictions on the future. AI currently looks like it will take over everything if you trust the hype, but right now we’re in a place where we don’t know the costs of AI for various reasons:

• All the services providing AI tools to us are heavily venture capital juiced or get free azure credits to run their stuff, because microsoft is involved.
• We’re already seeing some effects like search results being wastly better when scoped to “before AI”, but in general I don’t think we can currently quantify the negative results of the quantity and quality of AI generates stuff.
• Leaky abstractions. While there’s for sure the case to be made that “craft code” might be over I’d argue that has been the case long before AI. Look at the bazillion wordpress websites out there having been cobbled together to barely working and somehow someone was happy with that. That approach works until you need reliability, reproducability and potentially become liable by law to a certain standard. AI like a junior dev can provide input into certain decision making, but given it’s a machine it cannot be the decision maker anymore. Just like you wouldn’t want a junior dev to decide on e.g. your clustering mechanism. And making (good) decisions will always require expertise or things will eventually go wrong. Leaving to much to AI will eventually hurt.

Whether you’re pro or con, just learn it. Get proficient in the use of local, open-source LLMs.

Do you have a source for this claim?

In a perfect definitely world it would be that easy, but in ours it’s like being at least few years behind others.

As wrote BigTech companies used money and all their power to popularise their own products and subscriptions. Now everything is based on subscriptions. It’s not like in old, good times when you could simply buy a movie.

Now usually at most you can go to a cinema or pay for at least few subscriptions. Having in mind company practices the same is going to happen with us. By definition I would be crazy if by seeing same patterns I would expect different results.

If people would at least look at the patterns or follow old good practices like “follow the money”, so they would see what’s coming next then maybe we could do something about that, but so far things are most likely not going to change as this way is “easier” a mass … and if you are not with a mass you are against a mass which is just a straight way to become an outsider and then nobody care about “outsider feelings”.

If someone really need an example then look at American movies/series … Everyone there someday goes on the long road and if car stops working they take tools and they are fixing the issue.

Should we claim that there is no sense in investing into any car repair company without understanding the context? So what’s the context? On the long road you often don’t have mobile connection.

Therefore you have to change the state of your car to make it work up to the nearest city where someone would fix the issue. If you know this fact are you still scared? Of course not! Now think carefully … Who have interest in scaring you?

Or maybe you think it’s just “coincidence” that so people “randomly” made rebranding on their own, on something that by coincidence is destroying humanity in the movies? When was the last time you’ve named a cup of coffee like an alien’s juice?

Now think by coincidence millions of people are scared about a cup of coffee as it could be an alien’s juice and you can’t prove it’s not. How do you prove that the first coffee’s seeds were naturally created on Earth? You can’t? That’s a prove! Don’t you have a fear?

You could buy it, but you would never own it. You bought the ability to play it (privately) on your device - maybe a friends device - until the tape broke.
So I’m not sure what you mean when you say “buying a movie”.

Anyhow back to the topic.
For me it is like that little plastic bricks, I love building anything with them, I start by putting two bricks together, and than it works itself till the solution.
AI is like, “I got that problem, generate me a building instruction for that”, which take the whole fun out of it.

Like other people said, I loose tinkering and exploring and seeing different solutions (in my mind, in real).

Which brings me to the point, I will only start a new project if I can see it in my mind - AI takes that vision from me.

True, boilerplate gets less boring with it (regarding coding, writing corporate emails), AI will help you stay afloat, but it will never show you the beatifiul underwater world.

Tha big question for me is more like, if an repressive government is using that technology againts its citizens (or citizens of another state), is that morally bad? Compared to other technology?