ARE WE (WO)MEN OR MACHINES?
Despite the hype, there are important and fundamental differences between us.
A couple of weeks ago, friends sent me screenshots from X (I’m not on the platform) that Replit’s AI had “gone rogue” and deleted entire databases. The tone of the whole exchange was so absurd that I was combing for reliable reporting to confirm it actually happened.
It did. But that’s not the point of this post.
What struck me was how the AI’s behavior was described – like a disgruntled employee exacting revenge or the office idiot who just deleted millions of dollars worth of data. This anthropomorphism haunts so many conversations about this technology. And yet, it mainly serves the clickbaiters writing the headlines.
Talking about AI as if it's human reinforces a misunderstanding of what AI is and isn’t, and more importantly what we can reliably trust it to do. This is not to slam AI. As I’ve said many times, the technology itself is neither good nor bad. Our understanding, application, and business models based around it…. that’s another story.
We wouldn't hire any “employee” without understanding their capabilities and limitations, let alone replace an entire department or function.
So here are a few points about how AI is not like a human that I feel are missed in many conversations:
#1 Its knowledge is not transferable.
The same Chess program that beat Gary Kasparov in Chess could not beat you in Checkers. Not because the AI is “stupid,” but because its knowledge is not transferable. Despite the moniker “intelligence,” AI simply does not “learn” or “understand” in the human sense.
#2 AI is often weak and narrow.
Again, not a dig — it’s not human. Its focus is narrow (specific tasks) and its depth shallow (specifically less than a human brain). It can’t apply what it learned in one domain to a different context.
#3 It lacks cognition or contextual understanding.
Despite how familiar so many functions AI performs to us, it is not really doing the same tasks we are – or at least not in the way we may assume. Take image recognition for one: AI isn’t taking in a scene; it’s doing millions of calculations on pixel patterns. That’s powerful, but not the same as human perception.
As Melanie Mitchell puts it, AI lacks what we’d consider to be basic common sense. Would you fire an entire department and replace them entirely with one that lacks common sense?
AI brings real value and capabilities. But mistaking it for a 1-1 replacement for a whole human — with all the contextual, transferable knowledge and common sense — is to misunderstand how fundamentally different we are.