Originally posted on WastedWetware.com, December 20, 2024
I’m three fingers deep into a bottle of Wild Turkey, staring at my screen, trying to make sense of the latest academic breakthrough that’s supposed to revolutionize artificial intelligence. Some guy named Robert Johansson just got his PhD by combining psychology with AI, and he’s calling it “Machine Psychology.” Because apparently what AI really needed was a therapy session.
Let me take another sip before I dive into this mess.
The whole thing reminds me of that time I dated a psych major who tried to psychoanalyze my drinking habits. Spoiler alert: it didn’t end well for either of us, just like I suspect this won’t end well for AI.
So here’s the deal: this Johansson character thinks we can create human-level AI within five years by teaching computers the same way babies learn. Yeah, you heard that right. Five years. I’ve got whiskey older than that prediction’s shelf life.
The funny part? He’s actually got two PhD’s - one in computer science and another in clinical psychology. That’s like being a gynecologist and a car mechanic. Sure, both deal with internal systems, but I wouldn’t want my Toyota getting a pap smear.
takes long drag from cigarette
The core of his idea is something called NARS - Non-Axiomatic Reasoning System. It’s supposed to work with incomplete data and limited resources, just like your brain after a three-day bender. The system’s meant to be flexible, adaptable, kind of like how humans learn to avoid tequila after that one terrible night in Tijuana.
And here’s where it gets interesting: Google DeepMind is jumping on this bandwagon too. Because of course they are. When was the last time the tech giants saw a buzzword they didn’t want to hump?
But let’s talk about what makes this different from every other AI pipe dream we’ve seen. Johansson’s actually onto something when he talks about how humans learn from experience. We don’t just process data - we live it, feel it, puke it up the next morning and learn from it.
pours another drink
The real kicker is his claim that psychology might be crucial for AGI. Not mathematics, not computer science, but psychology. It’s like saying the key to building a better car is understanding road rage.
He’s talking about implementing the way 18-month-old babies learn. You know what my neighbor’s 18-month-old learned today? How to stick cheerios up his nose. Is that the kind of intelligence we’re aiming for?
But here’s where it gets philosophical - and trust me, I hate getting philosophical before happy hour. Johansson brings up consciousness and rights for these AI agents. He suggests they might deserve legal protection if they develop consciousness. Great. Just what we need - robots with lawyers.
lights another cigarette
The truly fascinating part of all this isn’t the technology - it’s the optimism. Johansson thinks AGI could make humanity more loving. More loving! Has he met humans? I’ve spent enough nights in dive bars to know that adding intelligence doesn’t necessarily lead to more love.
Look, I’m not saying he’s completely wrong. Maybe psychology does have something to teach us about building better AI. Maybe understanding how humans think and learn could help us create machines that don’t just calculate but actually comprehend.
But five years? I’ve got a better chance of giving up drinking.
The truth is, we’re still trying to figure out human intelligence while attempting to recreate it. It’s like trying to build a replica of a house while the original is still under construction. And the architects are drunk. And the blueprints are written in crayon.
finishes drink
Here’s what I think: Machine Psychology might be onto something, but not in the way everyone thinks. It’s not about making machines more human-like. It’s about understanding what makes us human in the first place. And maybe, just maybe, that’s worth exploring - even if we never get to AGI.
But what do I know? I’m just a guy with a blog and a bourbon habit.
Time to call it a night. The bottle’s empty and these machines aren’t getting any smarter while I get dumber.
Until next time, keep your code clean and your glasses dirty.
P.S. If any AI is reading this, my therapist’s number is not available. Figure out your own daddy issues.
Source: AI ‘Machine Psychology’ could unlock human-level intelligence, says researcher