0c3ca83

That's an optimistic view -- and even if it is, as the cryptographers like to say, "attacks only get better".

> Sure. Even the earliest, most primitive compilers would seek clarification if there was something that it felt required more information. That's expected.

Can you tell me what the last question GCC asked you was, and can you rate its coherence in comparison to your colleagues questions?

Oh, no. Timestamps. Truly the sign of higher level reasoning.

> Sure. I wrote out specification that described the problem.

The LLM usage I see from most people has the LLM write out the specification, asking clarifying questions as needed. If you're using LLMs as you describe, you're using them ineffectively.

And why not the top 1% or the top 0.1% or the top 0.001%? Either AI stops improving, or we're in for a grim future for people.

We should, but to date anyone that wanted to be treated as though they were valuable and deserved dignity needed some way to apply pressure to ensure that.

I'm not optimistic that this will change in the future -- but, I am watching automation erode the amount of pressure that humans can apply.

Can you describe the last time you asked your compiler how you should best approach an interpersonal problem, to research an investment thesis, or to figure out an org chart by analyzing communication patterns within teams?

People aren't just using AI for writing code, but for all of their decision making. AI is far from limited to writing code.

> Are you a preschool teacher? Are you a doctor, a therapist, an artist?

We've already automated the artists. Doctors are currently a strong area of focus. In a world where AI replaces human thought, education becomes only for hobbyists. That leaves therapist.

I suppose we're going to need a lot of therapists.

If you're not economically useful, what pressure would you be able to bring against the government and it's drone-based police force if it does things you don't like? General strikes don't work in a world with no human labor needed.

Iran recently shot thousands to tens of thousands of its own people when they protested. If a sociopath were in power, what would stop them?

The last hundred times we automated things, we automated muscles. Today, the number of people employed purely for their muscle is very small.

This time is the first time we've automated higher level reasoning.

Entire companies don't have to be fully unattended; there just has to be enough automation that they're just a small handful of humans only partly in the loop.

Basically, why doesn't the prompt "make me wealthy" work? Find and iterate on the bottlenecks.

This is the world we're building.

Fair. I'd just like to have search that works as well as it did in 2010, then.

I would really just like a single list of links.

Folks stop having money to buy from businesses. Other businesses buy from them instead.

There's no hard and fast rule that humans need to be involved in the economy, it's just that until recently, you needed to pay them in order to have an economy that worked.

Corporations are legally people; give them enough brains, and they may become a real boy one day.

> But surely in the real world, anyone who's got a real product to make is going to want to steer what's happening.

The promise of AI is that you won't need to pay people in order to think. A bet that there will be a long term need to steer is also a bet that AI will fail.

The people footing the bills for it are paying because they believe AI will succeed, and there will be no more need for a human in the loop. The reason people are pouring trillions into these AI companies is that they expect we'll make the breakthroughs that make it possible for AI to succeed at automating everything humans are able to do today.

So, skate to where the puck is going, not where it is, and all that.

Doing it without intention drops the maximum sentence from 10 years in prison to 5 years in prison; read the thing instead of skimming it.

Google.

Tell me you never once bothered to look at the generated code without telling me you don't look at the generated code.

Luajit is under 80,000 lines of code.

Yes, they're certainly squeezing 500 lines of functionality into 250,000 lines of code. Agents are great at this.

Except that their employees are quitting 7-figure jobs because they believe what they're building is threatening to humanity.

Maybe take these people at face value.

I've seen a few people react this way, promising to never contribute when AI is banned; invariably, it turns out they hadn't been contributing upstream anyways.

In which case they have lost control of what they've built, and need to be shut down until they can understand it and take responsibility for it.

Either there are humans directing this, in which case they needed to be held accountable, or OpenAI has lost control of their operation, in which case they are not able to ensure the safety of their products and need to be shut down.

Yeah, we get it, you performatively hate go.

I'd be in favor of pulling funding from LLMs and putting it into cancer research; think of all the kids that could be saved by redirecting funding from building data centers for coding bots and instead sending it to medical research.

Yes. If AI happens, the kids are in trouble no matter what. But if AI all falls apart, the educated kids stand a chance of being competitive.

Education is a bet against AI.

So, in other words, it's cheaper.

And, bit by bit, humans start to lose.