[GCA10]

I've seen variants of this in finance, where the strategy is called "outrunning your mistakes."

There's an art to building a rapid growth trajectory, cutting all kinds of corners to ramp up your customers/assets/users/whatever -- and then pirouetting into a bigger new role, while leaving someone else to stabilize your unstable creation.

Take it up a level, and the best practitioners are also very good at blaming their successors for messing up a good thing. Also very good at cultivating a certain subset of journalists who will keep burnishing the legend, never stepping outside the bubble.

Loosely related: does anyone in the brave new AI community have the patience and humility to try to solve the hallucinations problem? Because it seems as if Sam & Co. are quite determined to ignore it, deny it or minimize it, while continuing to push for breakneck growth of products with a very troubling design flaw.

[kevinventullo]

does anyone in the brave new AI community have the patience and humility to try to solve the hallucinations problem?

I think this is akin to asking if anyone in the software engineering community has the patience and humility to try to solve the software bugs problem.

[disgruntledphd2]

Nope, those are very different. Fundamentally the reason that LLMs hallucinate is because they are predicting the next most likely word. It would be expensive to find a way to remove outputs where P truth is low, but it's not completely implausible like removing software bugs.

[kevinventullo]

I suspect removing outputs where P(truth) is low is about as hard (and arguably a superset of) detecting code for which P(bug) is high.

As an aside, I think these are both roughly equivalent to detecting code for which P(halts) is high!

[voxl]

So more expressive type systems? Rust has made entire classes of bugs impossible, and you can do even better in Idris or some other dependent type theory. People don't do it because it's expensive and they're not trained it.

No it never eliminated all bugs, but if we're conflating the two problems you don't need to eliminate all hallucinations either, you just need to make them very unlikely, or a consequence of your bad specifications.

[throwaway778945]

Yeah, have worked for/been burned by a few of these types. It can be really demoralizing and sets a bad example for how to be successful. Over time I've decided that how I accomplish things matters just as much, if not more than, what I accomplish, and outrunning your mistakes is a terrible "how."

[itake]

This is extremely common in big tech careers.