> 1. All the things it can obviously do very well today
I'm curious what those things are.
At least to me, it isn't obvious that LLMs solve any of their many applications from the past year "very well". I worry about failures (hallucinations, misinterpretation of prompts, regurgitation of incorrect facts, violation of copyright, and more). I don't have a good sense of when they fail, how often this happens, or how to identify these failures in scenarios where I'm not a domain expert.
But maybe some subset of these are solved problems, or at least problems that are actively being worked on for the next generation of models.
Yes, there are many other kinds of AI. Stockfish is better at chess than any human. But when you start talking about emergent behavior from machine learning, the failure modes are much harder to reason about.
> At least to me, it isn't obvious that LLMs solve any of their many applications from the past year "very well". I worry about failures (hallucinations, misinterpretation of prompts, regurgitation of incorrect facts, violation of copyright, and more)
That’s a list of things that gets clicks in the popular press.
Some solutions I love:
Recording a video and creating a transcript from it. Then edit the transcript and the video gets edited.
Translating (dubbing) video and changing the lips of the speaker to match where they should be for the new audio.
Scanning invoices for mistakes (I know entire businesses build just in this one thing).
Understanding edge cases. So many things have some nice hard and fast rules that can fail, and an ML system of some sort can figure it out so the system can move on. I do this for processing SEC data, and soon for web scraping (like they change the html but visually it’s kinda the same, and the ai system can figure it, give me a new html selector, and back in business).
> That’s a list of things that gets clicks in the popular press.
Are you saying they're nonissues in practice? I agree that most of those points have shown up in the news, but they're also things that I have personally observed when interacting with LLMs.
Of your (and sibling commenters') cited use cases, I see a number of scenarios where AI is used to perform a quick first pass, and a human then refines that output (transcript generation, scanning invoices, iterating on transformations for syntax trees, etc). That's great that it works for you. My worry here is that you might heuristically observe that it worked perfectly 20 times in a row, then decide to remove that human check even as it admits more errors than is acceptable for your use case.
> Scanning invoices for mistakes
This is one of those cases where I would like to better understand the false negatives. If a human reviews the output, then okay, false positives are easy enough to override. But how bad is a false negative? Is it just unnecessary expenses to the company, or does it expose them to liability?
> Translating (dubbing) video and changing the lips of the speaker to match where they should be for the new audio.
This is useful in itself, but surely you too can see the potential for abuse? (This is literally putting words in someone else's mouth.)
> This is useful in itself, but surely you too can see the potential for abuse? (This is literally putting words in someone else's mouth.)
If I was a famous actor, I would demand it. I don’t want people hearing different voices for different movies. I want them to hear my voice. And I’d want it to be authentic. Seeing lips move to the wrong words does not help make a connection.
As for abuse, sure. Not sure anything has had worse abuse than database technology. There should definitely be an avenue for the government to shutdown any database instance anywhere (like California is doing with AI). I would have shutdown data broker databases long ago.
> This is one of those cases where I would like to better understand the false negatives. If a human reviews the output, then okay, false positives are easy enough to override. But how bad is a false negative? Is it just unnecessary expenses to the company, or does it expose them to liability?
In the companies I know about, these invoices requesting overpayment just got paid. So, worst case, it’s the same. But best case there is way way way more money to save than the cost of the service.
I find them pretty good at reasoning about tree structures that have a depth which I myself find difficult to navigate. For instance, I've been working with libcst (a syntax tree library) and I can say:
1. observe this refactor rule which I like
2. here's the starting code
3. here's the desired code
4. write me a refactor rule, in the style of 1, which transforms 2 into 3
It sometimes takes a few iterations where I show it a diff which highlights how its rule fails to construct 3 given 2, but it usually gets me the transformation I need much faster than I'd have done so by hand.
And once I'm done I have a rule which I can apply without the LLM in the loop and which is much more robust than something like a patch file (which often fail to apply for irrelevant reasons like whitespace or comments that have changed since I wrote the rule).
The key is to find cases, like this one, where you can sort of encircle the problem with context from multiple sides, one of which works as a pass/fail indicator. Hallucinations happen, but you use that indicator to ensure that they get retried with the failure text as correcting context.
It helps to design your code so that those context pieces stay small and reasonably self-describing without taking a foray deep into the dependencies, but then that's just a good idea anyway.
> But when you start talking about emergent behavior from machine learning, the failure modes are much harder to reason about.
Sure, this is basically why so many are concerned AI might kill us all:
Lots of observed emergent phenomena that are not expected (basically everything ChatGPT can do given it was trained on next token prediction), as a result of doing exactly what we said instead of what we meant (all computer bugs ever), doing it so hard that something breaks (Goodhart's law), doing it so fast that humans can't respond (stock market flash-crashes, many robotic control systems), and being so capable in smaller scale tests that they tempt people to let go of the metaphorical steering wheel (the lawyers citing ChatGPT, but also previously a T-shirt company that dictionary merged verbs into "keep calm and …" without checking, and even more previously either Amazon or eBay dictionary merging nouns into "${x}: buy it cheap on {whichever site it was}" with nouns including "plutonium" and "slaves").
If people had a good model for the AI, it wouldn't be a problem, we'd simply use them only for what they are good at and nothing else.
I'm curious what those things are.
At least to me, it isn't obvious that LLMs solve any of their many applications from the past year "very well". I worry about failures (hallucinations, misinterpretation of prompts, regurgitation of incorrect facts, violation of copyright, and more). I don't have a good sense of when they fail, how often this happens, or how to identify these failures in scenarios where I'm not a domain expert.
But maybe some subset of these are solved problems, or at least problems that are actively being worked on for the next generation of models.
Yes, there are many other kinds of AI. Stockfish is better at chess than any human. But when you start talking about emergent behavior from machine learning, the failure modes are much harder to reason about.