Hacker News Reader

Comment by MengerSponge | Hacker News Reader

A COMPUTER CAN NEVER BE HELD ACCOUNTABLE THEREFORE A COMPUTER MUST NEVER MAKE A MANAGEMENT DECISION.

While I have great respect for this piece of IBM literature, I will also mention that most humans are not held accountable for management decisions, so I suppose this idea was for a more just world that does not exist.

I'd say that the fix then is in creating a more just world where leaders are held accountable than to hand it off to something that, by its very nature, cannot be held accountable.

human CAN and computer CAN NEVER

Accountability is perhaps irrelevant is my point. You can turn off a computer, you can turn off a human. Is that accountability? Accountability only exists if there are consequences, and those consequences matter. What does it mean for them to "matter"?

If accountability is taking ownership for mistakes and correcting for improved future outcomes, certainly, I trust the computer more than the human. We are never running out of humans incurring harm within suboptimal systems that continue to allow it.

Let’s assume we live in a hypothetical sane society, and company owners and/or directors are responsible for their actions through this entity. When they decide to delegate management to an LLM, wouldn’t they be held accountable for whatever decisions it makes?

Computers are more accountable. You just pull the plug, wipe the system.

Executives, in contrast, require option strike resets and golden parachutes, no accountability.

Neither will tell you they erred or experience contrition, so at a moral level there may well be some equivalency. :D

>> Computers are more accountable. You just pull the plug, wipe the system.

I think you are anthropomorphizing here. How does a computer feel when unplugged ? How would a computer take responsibility for its' actions ?

I wonder if that quote is still applicable to systems that are hardwired to learn from decision outcomes and new information.

LLMs do not learn as they go in the same way people do. People's brains are plastic and immediately adapt to new information but for LLMs:

1. Past decisions and outcomes get into the context window, but that hasn't actually updated any model weights.

2. Your interaction possible eventually gets into the training data for a future LLM. But this is incredibly diluted form of learning.

What (or who) would have been responsible for the Holodomor if it had been caused by an automated system instead of deliberate human action?

Management is already never held accountable, so replacing them is a net benefit.

Reply on news.ycombinator.com