With great (AI) power comes great (human) responsibility
The new Spiderman film is out and all I can think of is how appropriate it is to apply some OG Spiderman principles from uncle Ben to our use of AI.
Honestly, I'm tired of AI.
Every day we get hundreds of emails, Slack messages, documents, pull requests, you name it, that are 100% AI generated.
When did we stop thinking and start over-relying on AI? When did everything become a five page spiel full of jargon?
These days it's no longer em-dashes or capitalising every word that gets to me. The vocabulary alone is so off-putting, it takes reading a single sentence to spot that it’s completely AI generated.
Just this week I received a customer complaint saying they weren’t able to integrate with a platform because the feature isn’t supported. Turns out, the platform has the capability just fine, but they were so convinced by Claude that they didn't bother to verify what turned out to be a hallucination.
Then last week I read a company policy that was obviously AI generated, containing irrelevant clauses and emojis in every clause title, to make it…fun?
And then the every day issue where I'm reading a pull request with an eight paragraph description full of words like cutover, leaky, exposure, chunk, mint, cross-layer, and many others that we would never use in conversation when describing a code change to a colleague.
Outsourcing your work to AI is one thing, but outsourcing your judgment, that’s arguably unacceptable. There are people that write entire emails in which they refer to themselves in third person, and they send it without even proofreading. John wrote “John will be working on x, while Paras works on y”. Okay John, let’s pretend that Claude didn’t draft that email.
It’s easy to one-shot an email draft with AI, and send it off to the recipient, but little do we consider that it simply pushes the burden onto the recipient who now needs to read your AI generated two page email that could have been 5-6 sentences.
It doesn’t take amazing pattern recognition to spot AI generated text anymore, they all sound the same, they all overuse the same word, everyone on LinkedIn is always “delighted”.
It’s not merely annoying, it’s making us dumber.
See what I did there? Did it make you cringe? Because even though I wrote it, I cringed. There’s also the issue of feeling like something is AI written even when it’s not, to the point where a lot of people I know feel uncomfortable using em-dashes today, even though they were using them pre-AI.
Niklas Gruhn recently coined the phrase “meat proxy” for people who blindly copy and paste the output from AI to others, and I think it’s blatantly true. Scary as well, because half the people I know are turning into meat proxies.
Have we considered how our usage of AI impacts the human brain? The brain is a muscle and the less you use it, the more it declines. In order to keep the decision-making parts of our brain sharp, especially the prefrontal cortex, we have to exercise by making decisions ourselves.
Choosing from Claude’s multiple choice options is not quite the same. A recent study titled “Your Brain on ChatGPT” found that LLM users struggled to accurately quote their own work, simply because of their reduced cognitive activity and lower engagement with the subject at hand.
While everyone is so focused on the next frontier model and how “low cost” it is, we should instead dig deeper into its true, hidden cost, that is not advertised.
Sam Altman recently called out young people overrelying on ChatGPT and this alone should have us rethinking our dependence on AI, in my opinion. The same people selling you the product are telling you that they’re worried about its use. Capitalism isn’t going to stop them selling it to us, but it’s our own responsibility to protect ourselves from its harmful effects, not very different from alcohol, drugs, gambling, etc.
A couple of years ago when ChatGPT became popular, we said “AI will never be able to …” whereas these days it feels like AI is doing the heavy lifting in everyone’s day-to-day jobs. Let’s not forget that AI was developed to help us, not to make us dumber or be the cause of our cognitive decline, even shorter attention span post tiktok/reels, and overall inability to do anything ourselves without this crutch.
Bottom line, coming back to Spiderman, Uncle Ben coined it first, “with great power comes great responsibility” and that still applies today in the world of AI. Use AI. Don’t write a single line of code yourself, fair enough, why should you? When AI can do it quicker and better! But, use your judgement. You drive the AI, and not the other way round, and where you are able, you do the work, yourself. Write that company policy first draft and have AI nitpick the details, the legal validity, etc. Write that email and ask AI to have a look over it, to ensure it lands how you want it to. When AI gives you an answer, fact check it. Do the work.
The world is filling up with AI slop, and very soon we will have the great reset, where everyone is going to spend time rewriting, redesigning, removing, repairing the slop. When that time comes, we need capable human brains to take on the work. Get ready!