Everyone today is, and I say this begrudgingly, rightfully concerned about the new Mythos model announced by Anthropic this past week.

Basically this model is quite advanced to the point where it is discovering bugs and “zero day” (aka unknown) issues on 30 year old tooling that no one has touched for years.

Which is alarming.

I suppose.

At least I know my company is currently freaking out about it. As is nearly every other company inundated with massive amounts of tech debt who have willfully turned a blind eye to it for literal years. Suddenly we all have only six weeks to get ourselves prepared for this (dystopian?) future.

Present freaking-out excluded, we’ve also managed to do a whole lot in the world of computers over the last 30 years. No one was the wiser as to these long-dormant bugs.

Say Anthropic never created this model, either intentionally or not. Would that actually be a problem in our world today?

Our AI overloads would definitely use the same excuse “if not us then someone else!” so we should feel grateful that they’re telling us all ahead of its release.

It honestly sounds a lot like the ever-looming quantum future of computing.

Quantum computing, the little if anything I know about it, is essentially computing that does not rely on typical zeros and ones when it comes to representing data in memory registers. Imagine if you had a computer that could handle anything from zero to, say, a hundred within each bit in some way.

Today’s cryptography just wouldn’t keep up in that world. So things like 2048 bit encryption really have no chance against a potential threat like that. At least as I understand it.

Some of this feels similarly.

Progress for the sake of progress isn’t actually progress for humankind.

Not always at least.

My opinion here is more political actually. If we never allowed a company to quite literally steal the world’s data to train their models, then perhaps they’d never have created Skynet, I mean, Mythos in the first place.

Would we be worse off for it? Probably not.

Meanwhile, in the trenches of my day to day job…

I had a pull request to review today that had over 351 comments on it, 35 different commits, modifying only 3 files, nearly all of which was generated by Claude. A huge portion of it was rewriting slight variations of other code that already existed in the same repository. Or better yet, redefining standard library code, a known issue in a lot of these actions cited by others.

This is proposed core build CI infrastructure code!

So on one side, we have these ridiculously overpowered and world-ending models put out by companies worth trillions of dollars…

And on the other hand, we have a bunch of AI slop becoming ever-increasingly relied on by lazy engineers who are squeezed to death by management who really have no idea how else to consider themselves successful in their jobs.

What a world we live in.

I went earlier on a walk with my dog and saw people out on their evening walks. They were on their bikes riding home. People were active, enjoying the beautiful weather outside today.

If we accidentally (or purposely, slowly) sign on to having AI run our bank accounts and suddenly we onboard Skynet to the world around us, what’s the point of it all?

We’re human, we should be doing human things. Isn’t that the point of technology, to better add to our lives?

If instead of usurp the beauty that is life with AI models that start to make autonomous decisions and we get lazier and lazier in letting it do all the important work for us, we rob ourselves of a better life. We instead place all control and money into the hands of a few who won’t know how to utilize it when that day comes.

Or worse yet, they’ll simply weaponize it against us. Imagine rioters with pitchforks charging the nearest data center, but then seeing others fight against the action because, well, that’s their banking and picture data too!

I imagine it to be somewhere between Wall-E and the Terminator movies.

Realistically, this is where government oversight and anti-trust laws should help us balance new features and technology with the desire to be human.

So again…

Just because we can doesn’t mean we should.