Sabitlenmiş Tweet
AI Skeptics Are Right About the Wrong Thing
The issue is not that developers doubt AI. The problem is that many of them are doubting the wrong part.
Tonight at dinner, a developer told me AI is basically autocomplete with a giant electricity bill…
It made me laugh, because it is a good line.
It is also, in a technical sense, not completely wrong. Large language models predict tokens. They do not have beliefs, intent, responsibility, taste, or the understanding a senior engineer has when a production system breaks at 2 a.m.
Then the cost part….
The International Energy Agency says data centers used around 415 TWh of electricity in 2024 and could more than double to around 945 TWh by 2030, with AI as one of the main drivers.
And also right about quality.
Stack Overflow’s 2025 Developer Survey found that 66% of developers using AI tools are frustrated by answers that are “almost right, but not quite.” Another 45% say debugging AI-generated code can take more time.
If the point is that AI is expensive, unreliable and overhyped, I agree.
Anyone building with it should do so.
AI can hallucinate or invent APIs. It can hide bad assumptions behind fluent language or make bad developers faster at producing bad code.
It doesn’t replace taste, architecture, or the responsibility to verify what ships.
But where I think the skeptic argument becomes too small is in the jump from:
AI is prediction
To
AI is not useful.
Because a lot of valuable work starts with prediction.
Devs recognize patterns from old bugs. Designers feel when something is off before they can explain why. CEOs make bets from incomplete information. Writers know a sentence is wrong before they know the fix.
We don’t call that autocomplete, because it happens inside a person.
Of course human thinking is not the same as machine prediction. That distinction matters.
But it does not make the tool irrelevant.
—
The interesting part of AI is not that it thinks like us.
More so is the part that it gives people a new surface to think against.
It can compress messy context. It can show different angles. It can draft the bad version so you can react to it. It can challenge assumptions, suggest a test, explain a tradeoff, or connect two ideas you would not have put next to each other yet.
It’s not intelligence in the human sense.
Call it leverage.
And developers are already acting like it.
Stack Overflow says 84% of developers are using or planning to use AI tools in their workflow, and 51% of professional developers use them daily. GitHub’s Copilot research found that developers completed a coding task 55.8% faster with Copilot than without it.
That does not mean AI makes everyone better.
It means the job is changing.
For a long time, the core developer question was:
Can you write the code?
That still matters.
But increasingly, the more important question is:
Can you define the problem clearly, steer the system, inspect the output, catch the subtle mistakes, and know when the machine is confidently wrong?
That is a different craft.
Some AI skepticism is technical, and it should be. The field needs people who ask hard questions about cost, quality, reliability, security, and whether the output is actually better than what we had before.
But some of the skepticism is emotional too.
If you spent years mastering code the hard way, it’suncomfortable to watch someone move faster with tools you do not fully trust.
That does not make the skepticism invalid.
It makes the transition human.
I do not think the future is people typing prompts into chatbots forever.
The interface will probably get thinner: IDEs, agents, voice, ambient context, maybe eventually brain-computer interfaces.
Less typing. More intent. More execution.
So I do not think the real question is whether AI “thinks” like us.
The better question:
What happens when people who know how to think start using machines that can execute at scale?
That is where this becomes interesting.
English
