These Damn ‘Thinking Machines’ and Their ‘Consistency’

Another day, another news item about some startup with billions of dollars and a gaggle of ‘all-star’ researchers trying to fix what ain’t broken. Now they’re calling themselves ‘Thinking Machines Lab’ – ‘Thinking Machines’! We had thinking machines back in my day, they were called computers, and they did what you told ‘em to do. You type in a command, it executes. Simple. Elegant. None of this ‘inconsistency’ nonsense they’re prattling on about. It’s like they’ve never heard of a subroutine. Or a compiler. Or, for the love of God, a single, solitary IF-THEN statement.

The Myth of AI ‘Improvement’

They talk about ‘improving AI models.’ Improving them? What are they improving? The ability to hallucinate even more convincing nonsense? To generate pictures of cats with six legs more consistently? This Mira Murati, with her two billion dollars in ‘seed funding’ – seed funding for what? To plant a garden of digital weeds that will choke out any semblance of logical thought? Back when I was at IBM, if we had two billion dollars, we’d build a mainframe so powerful it could calculate the trajectory of a fly farting on the moon, and it would do it the same way every single time. Consistently. Reliably. Not this wobbly, wishy-washy ‘it might do this, it might do that’ garbage.

The Good Old Days of Predictability

I remember the glory days. You’d sit down at a terminal, green text on a black screen, and you knew where you stood. You gave it an instruction, it gave you a result. Predictable. Like a well-oiled machine, because it was a well-oiled machine! Now, these ‘AI models’ are like trying to herd a flock of hyperactive, opinionated pigeons. They’re supposed to be ‘thinking,’ but all they seem to do is make things up. And now, the big breakthrough is making them make things up consistently? What kind of progress is that? It’s like celebrating a faulty printer that prints the same wrong page over and over again. It’s not a feature, it’s a bug!

The Real Problem: Too Many Choices

The real problem, if you ask me, is that these newfangled systems have too many choices. They’re trying to ‘understand’ context, ‘interpret’ intent. Nonsense! A computer’s job isn’t to interpret; it’s to execute. Give it a clear instruction, and it should follow it. This ‘nuance’ and ‘creativity’ they’re after? That’s what humans are for! We’ve got minds, we’ve got emotions, we’ve got the ability to actually think. These machines are supposed to be tools, not digital poets. And when your tools start freelancing, that’s when you know you’ve gone too far.

My Simple Solution

My solution is simple: get back to basics. No more ‘neural networks’ that nobody can explain. No more ‘deep learning’ that just sounds like a fancy way of saying ‘we threw a lot of data at it and hoped for the best.’ Give me a command line. Give me a compiler. Give me a machine that does exactly what I tell it to do, when I tell it to do it, and does it the same damn way every single time. That’s consistency. That’s progress. Anything else is just making things more complicated for no good reason. These kids and their fancy phones and their ‘thinking machines’ – they’re all just overcomplicating things. Probably too busy looking at their Instagrams to actually learn how to code properly. Pfft. Back to my BASIC manual. At least that makes sense.