Except when they "extract" something that wasn't in the source. And now what, assuming you can even detect the tainted data at all?

How do you fix that, when the process is literally "we throw an illegible blob at it and data comes out"? This is not even GIGO, this is "anything in, synthetic garbage out"

> Except when they "extract" something that wasn't in the source. And now what, assuming you can even detect the tainted data at all?

You gotta watch for that for sure but no that's not a issue we worry about anymore, at least not for how we're using it for here. The text that's being extracted from is not a "BLOB". It's plain text at that point and of a certain, expected kind so that makes it easier. In general, the more isolated and specific the use case, the bigger the chances of the whole thing working end to end. Open ended chat is just a disaster. Operating on a narrow set of expectations. Much more successful.

> Except when they "extract" something that wasn't in the source. And now what, assuming you can even detect the tainted data at all?

I mean, this is much less common than people make it out to be. Assuming that the context is there it's doable to run a bunch of calls and take the majority vote. It's not trivial but this is definitely doable.

I really don’t think that’s doable because why do you the majority output is correct? It’s just as likely to be a hallucination.

If he problem is the system has no concept of correctness or world model.

Assuming that hallucinationd are relatively random it's true. I do believe that they happen less often when you feed the model decent context though.

I mean, it is obvious for a human inspecting the one specific input and output sample, but how do you do this at scale? (Spoiler: cross your fingers and hope, that's how)

In which case...what good is a model that predicts semi-randomly? Oh.

("But it works - when it works" is a tautology, not a useful model)

What does "semi-random" even mean? Are humans not "semi-random" in the same sense?

That is, literally, faith-based business management. "We suck, sure - but wait, a miracle will SURELY happen in version 5. Or 6. Or 789. It will happen eventually, have faith and shovel money our way."

By then, the startup will have folded, and the C-levels will have moved on to the next Idée Du Jour.

- terminally online

- DLC

- IRL

As in, "things that happened before my Xth year of life are Normal, nay, Traditional even. Everything afterwards is Ephemeral, and possibly Heresy as well."

Alas! All is ephemeral.

HTAs were equal parts great (for their time) and terrible (plus they stuck around until IE died): a bog-standard webpage, but IE-only and with local-user process execution powers. Also, the local persistence was...problematic.

Only similar in the vaguest sense.

I am afraid that you underestimate the power of organic, human stupidity.

How I envision this to work:

- vibe code a thing

- it doesn't work, but it was CHEAP

- hire someone "to fix it"

- actually means "rewrite most of it, but cheaper"

- also means "become the scapegoat"

To a developer, it offers most of the work for very little payment. Not an enticing proposition.

(Yes, I've been burned by various "it's 90% done, we just need a few fixes", and that was BEFORE the current slop era)

Data at rest, also? Meanwhile, same govt. invests in LLMs slurping power by the megawatt, but surely that is an unproblem.

IDK if any pulse dialing phones (as opposed to DTMF) are still capable of reaching the phone network at large, but I'd imagine that's the use case.