DEV Community

Cover image for But At What Cost?
Alex Hyett
Alex Hyett

Posted on Originally published at alexhyett.com on

But At What Cost?

As a race, humans are really bad at doing cost benefit analysis. We might try and weigh up the pros and cons but we all have a tendency to exaggerate the pros if they are beneficial to us and downplay the cons that we can't directly see. This is also the case if the downsides are not immediate but will affect us at some point in the future.

We see this pattern throughout history. We find something useful or pleasurable and later discover that it is harmful or unethical, and then continue doing it for another 50 years.

Asbestos #

Take asbestos for example. It has been used for hundreds of years, it is dirt cheap, strong and fire resistant. So we used it everywhere in tiles, in furniture and even in clothing.

The first documented death from asbestos was in 1906. In 1918 insurers stopped covering asbestos workers due to the known health risks.

It took until 1999 to completely ban it in the UK and until 2024 in the US. We knew it was killing people but we kept using it because it was cheap and it didn't directly affect those benefiting from the cost savings.

Smoking #

People have smoked tobacco for years and it was introduced to Europe by Columbus in 1492. Smoking was seen as cool, and a way to help you relax. It was given to soldiers in WWI and WWII to help with the stresses of war.

As early as 1912 we saw evidence that smoking caused lung cancer. In the 1940s it was common to see adverts from tobacco companies featuring doctors saying their brand was healthier than others. It wasn't until the study in 1950 that people started taking the link seriously.

Unlike asbestos though, cigarettes haven't been banned but at least smoking in public places has (in a lot of countries) and it is becoming less popular over time. Still in some countries as much as 30% of the population still smoke.

Meat #

The same is true for the meat industry. I think we all know that our obsession with eating meat is damaging the environment and the way the animals are treated is ethically wrong. Livestock production counts for 14-18% of greenhouse gas emissions with the majority coming from cows.

I was vegan for a few years but I have since reintroduced meat back into my diet at a reduced amount. I don't consume dairy products and I very rarely eat beef. It is mostly just chicken, fish, and some pork.

We have meat free dinners 2-3 times a week. I switched back to eating meat mostly for health and convenience. I am not the greatest cook and found myself resorting more and more to processed vegan alternatives than whole foods which wasn't great.

I always buy free range eggs and chicken but I know that "free range" isn't what we think it is and they still spend most of their lives in cramped conditions. It is easy to ignore the negatives when they don't affect us directly.

AI #

You might have guessed where this was heading.

The same pattern can be seen with people's AI usage. AI allows us to generate code, images, text, and video. In the past I have used AI to generate some useful personal tools and funny images. AI allows people to create things they wouldn't otherwise have the skills or time to create.

As with all the above examples, few people consider the negative effects and only see how it benefits them.

As I write this, in my area of the UK, it hasn't rained properly for nearly 2 months. The majority of the country is in a drought and all the fields look dead. The rest of Europe is suffering a similar fate with wildfires cropping up everywhere.

Despite this, the UK still plans on building over 100 data centres over the next few years due to the popularity of AI. It is difficult to get direct numbers for how much water is used for these data centres as it depends on how they are cooled. Those with evaporative cooling towers use millions of gallons of water per month whereas most are now built with closed loop systems which use less but still require a lot of water on the initial fill.

Even if the data centres aren't using much water they still require a huge amount of electricity (20 MW to 1 GW). The proposed plans are likely to double the UK's energy usage. Some of this power will come from renewable sources but the increase strain will cause more to come from fossil fuels and nuclear which contribute to greenhouse gases and water usage.

I have only focused on the environmental impacts here and I haven't even touched on cognitive decline, the destroying of books or the ethical or legal implications of the training data used.

I am not going to tell you to stop using AI but just to consider whether the benefits for yourself are worth the cost. Do you need to use AI for every little code change or could some of it be done manually?

Local AI doesn't solve everything as it still uses more power and is still trained on illegally obtained datasets. It doesn't however have the same environmental impact or result in as much cognitive decline (due to lack of capability) as cloud based models.

I plan on doing a detailed analysis of the pros and cons of AI in the future. Personally for me the costs outweigh the benefits so I won't be using AI anymore.


❤️ Picks of the Week #

🛠️ Tools #

Building and shipping Mac and iOS apps without opening Xcode - I wish I found this when I was developing HabitTed. I did most of the coding in Zed and switched back to XCode to build and debug.

The git history command - I knew about git history of course but didn't know there were more commands that could be used with it.

A Git hook to prevent committing directly to main - There was a VS Code extension that I used to use for this but this is a better way.

💻 Programming #

Your 'app' could have been a webpage (so I fixed it for you) - I love it when people reverse engineer things and get equally annoyed by everything being an app.

🔒 Security & Privacy #

Every new car sold in the European Union must include a driver monitoring camera - Well this is disturbing. Is anyone else sick of living in a Black Mirror episode?

Sanitize YouTube Share Links - More tracking we need to try and avoid.

The Great VPN Humiliation: Why the Government Actually Backed Down - I am glad the UK government gave up trying to ban VPNs.

🎨 Design & UX #

Collection of Digital Clock Designs - These are really cool and creative, although I don't know if I could use some of these on a daily basis.

🤖 AI #

Our failure to mediate the mess - As a developer I am pro-tech. I grew up in an era where each innovation made our lives better with minimal impact and zero privacy invasions. As I have said above, we shouldn't put technological progress above everything else though.

Punch yourself in the face with reality - This is definitely true. People made useless software before and they are just making more with AI.

The Prompt-Wait-Evaluate Loop: How AI Kills Flow Without You Noticing - When I had to use AI for work it was the context switching that bothered me the most.

The LLM Critics Are Right. I Use LLMs Anyway - "Cognitive dissonance" describes AI perfectly. It is the same as an animal lover still eating meat. Many developers know and agree with all the downsides of AI but continue to use it anyway.

Decoy Font - I am not sure this is the best way to combat AI web-scraping.

🎮 Gaming #

18 Words - My wife loves word games and we both found this quite fun. I must admit I couldn't get all 18 in the time though.


💬 Quote of the Week #

Our new challenge as little snowflake humans will be to prove we aren't language models.

From the article The Expanding Dark Forest and Generative AI by Maggie Appleton.

Top comments (0)