A phenomenon I've noticed recently is people trying to occupy some untenable middleground wrt to the use of systems sold as "AI" -- this is a position where people try to recognize the harms of this tech but also hold space for "responsible" or "ethical" use.
π§΅>>
When someone is trying to hold this untenable position, a few things tend to come up (not everytime, not everyone):
1- Defensiveness. People read criticism of the systems and proposed uses of the systems as accusations that users are "bad people". Thus a criticism of the tech lands as criticism of the user, and tensions flare.
>>
4- Tone policing. People who are trying to occupy that uncomfortable, untenable space will claim that clear statements of harms/strong principles against use of these systems will "turn others away" as if the centrists are the ones actually pushing for more ethical practice.
But this "other people won't listen" remark I think is really a way of saying "This makes me uncomfortable" while trying to claim to be on the right side of history at the same time.
>>
OCR can be a useful tool for many research projects! OCR is also the kind of technology that gets better with better language models, i.e. more fine-grained models of which word(parts) go where. That has been true since before "genAI" and will be true after.
Just because you can use the synthetic media extruding machines to approximate the task of OCR, however, doesn't mean that that task can or should be used to justify the use of "genAI" in research.
>>
I think another important step is a values examination. What is important to you? How are those values supported or not by entering the discourse in a way that holds space for OpenAI/Anthropic/Google/Meta and all the other actors in this massive push to shove "AI" into every part of our lives as "not all bad"?
What are your research goals, what do you value about participating in scholarship, how can you meet those goals/act in accordance with those values and what obstacles are in your way?
>>
Part of what makes that middle ground untenable and uncomfortable, I think, is that it requires carrying water for these clearly bad actors. You can set that bucket down and step out onto firmer ground.
This does require going against the mainstream, but that gets easier when a) you find you're not alone and b) you see how much of mainstream opinion on this is actually the result of marketing.
/fin for now
@xvf17 @emilymbender THIS IS SO UPSETTING. I am one of those folks advocating for using certain types of machine learning responsibly, like "identify the patients with cancer" which ML tools can do better than humans, and which were trained using ethical data sources.
But the LLM field is just a cesspool. I can *see* use cases for it, but holy shit fruit of the corrupt root system with that training data.
@SomeVeganCheeseIsOk
Oh no dude. This prof thinks you are an asshole then.
Read what she wrote again. She is full of fallacies herself.
@xvf17 @emilymbender
@xvf17
As far as I know AI was coined as simulated intelligence. I think LLMs are exactly that.
Ai is a term that we used to describe a bunch of systems. Chess engines, opponents in video games,... Now LLMs. Perhaps someday when we have even better systems we will call them ai, but for now ai seems to fit LLMs completely.
@Noisecolor @xvf17 @SomeVeganCheeseIsOk @emilymbender AI is literally the abbreviation of *artificial* intelligence. Not simulated intelligence. That feels like a real "moving the goalposts" moment.
AI as it has historically/originally been used is referencing a legitimate, non-organic intelligence. LLMs are smoke and mirrors, when it comes to intelligence. There is no cognition, no understand.
Personally, I prefer GenAI over scare quotes, but both are more accurate than LLM = AI
@theadhocracy
Not at all. You can freely browse the web or ask a chat who coined the term and why. It will be an interesting read I can assure you and you will understand that ai is not meant to refer to a specific human level intelligence like you mean. It wouldn't make sense at that time anyway.
@Noisecolor @xvf17 @SomeVeganCheeseIsOk @emilymbender Okay, literally my first reult:
"The term "artificial intelligence" was coined in 1956 at Dartmouth College by John McCarthy and colleagues to describe the new field of creating machines that could think and learn like humans."
So, it *means* "artificial", not "simulated". It covers inorganic "machines". And it references *thinking* and *learning*.
We can argue semantics on learning, but an LLM ticks 0 of those boxes.
@Noisecolor @xvf17 @SomeVeganCheeseIsOk @emilymbender And just to head off "learning", yes genAI has roots on "machine learning". But the output, the LLM, does not learn "like a human". Some may have reinforcement algorithms, but all that's fundamentally doing is a loop: take the new data, run your previous "training" algorithm, update your output.
Humans learn through understanding and cognition. An LLM does not. Otherwise they'd be able to do simple calculus before they could write essays.
@theadhocracy
"Doesn't learn like a human": True. Doesn't matter. Learning is adapting based on data. LLMs do that. That's the definition.
"Just a loop": Human brains are also "just" electrical impulses. Mechanism doesn't negate capability.
"Calculus vs. Essays": Outdated. Modern models handle both. Training order is not intelligence limit.
@Noisecolor @xvf17 @SomeVeganCheeseIsOk @emilymbender Yeah if you're just going to redefine words, then I guess there go possible way to help you understand why people use specific terms to mean things...
Like, I get that language evolves, but this is absurd π "Learning is adapting based on data", sure if you want to argue that a model "learns", but that's not what anyone means or how any dictionary defines it.
Learning implies knowledge and understanding. And LLM cannot understand.
@SomeVeganCheeseIsOk @Noisecolor @xvf17 @emilymbender I can agree with that, but the term in this case was being used in a definition, specifically "machines that can learn like a human". That's why I was referencing it.
I did explicitly add a comment about how "learning" can be applied to algorithmic behaviour (ML etc.), but 1) that isn't the way the term is used in the definition of AI (scope is narrower), and 2) I don't agree that's enough to make the leap for LLMs to *intelligence*
@SomeVeganCheeseIsOk @Noisecolor @xvf17 @emilymbender But yeah, in this instance, I still think its a real stretch to claim that GenAI models **learn**. They may be born from reinforcement algorithms, and we may have adopted "learning" to describe aspects of that behaviour, but imo that is a marketing term.
Biological algorithms actually *learn* and modify their behaviour without oversight. So I'm not saying that ML is never learning. Just that I disagree that such a broad redefinition is okay
@theadhocracy
You are twisting and turning and doing mental gymnastics in order to get even an imaginary straw in which you can maybe be right.
While we have called ai romba robots for years without any issues form anyone.
And now some people developed a big problem with lmms. A very specific group of people that hate ai. Do you see it? So you see something not quite right with this situation?
@Noisecolor @SomeVeganCheeseIsOk @xvf17 @emilymbender Speaking of strawmen... Yes, lots of people mocked and were openly concerned about using "AI" as marketing for smart home devices and tools like Roombas. Those are also not AI, no one in the industry thinks so, there's no gotcha here.
Plus, one of the big distinctions is that the general public didn't fall for it in the same way, partially because a Roomba can't self-reinforce a delusional loop with its user, making them think its sentient.
@Noisecolor @SomeVeganCheeseIsOk @xvf17 @emilymbender Simply saying "we've called other things AI before, even when they clearly weren't" isn't a good argument, it's literally the point we're all trying to make.
Is an LLM a "more advanced" form of software than the path-finding in a Roomba? Yes. Do both have capacity to capture data and update their output over time? Yes.
Are either of them "intelligent"? No. [cont.]
@Noisecolor @SomeVeganCheeseIsOk @xvf17 @emilymbender What do I mean be "intelligent"? I don't mean the laymen definition in this instance, but the scientific definition.
The definition we use when discussing whether another species is "intelligent". You could call this "higher order intelligence"; I don't love that term.
The point is, if you class an LLM as "intelligent" in that way, you broaden the definition to every form of life. Even an amoeba "learns" in this manner. [cont.]
@Noisecolor @SomeVeganCheeseIsOk @xvf17 @emilymbender But, to return to your own argument, that isn't what the term "AI" was meant for when it was coined. It was explicitly talking about machines that learned **like humans**. That reasoned. That *understood*.
A Roomba no more understands what a chair is when its avoiding its leg as an LLM understands why there aren't four Rs in strawberry. Both have a form of data memory. Neither have *understanding*. Therefore, neither are AI.
@theadhocracy
I think this exchange has ran it's course. You clearly have a very subjective and narrow definition of ai and intelligence in general and you know what. Good for you.
It's not shared by almost no people and certainly almost no experts. You know who shares it, only the anti ai crowd. Strange?
I was hoping for an inkling of an open mind but I see now that's not there
Have a nice day. Take it easy.
@SomeVeganCheeseIsOk @Noisecolor @xvf17 @emilymbender To be clear: not entirely what I meant. I have no issues with the umbrella term "machine learning"; I actually think it works pretty well.
I am specifically talking about GenAI models/LLMs. Again, I think this is a case of linguistic drift. Are they born of ML models? Yes. Do they themselves **learn**. I think that's arguable.
They reincorporate additional data over time. But the core rules don't adapt (caveat incoming)...
@SomeVeganCheeseIsOk @Noisecolor @xvf17 @emilymbender ... unless they're in some kind of feedback loop, running via multiple layers of models, some of which then manipulate those they have greater write level over. The "shepherd and sheep" model stuff.
But that isn't, again, what most people are talking about. An Agent "learning" your habits isn't the same thing. The input/output model isn't adapting, it's just getting more input.
I think we broadly agree anyway π
@Noisecolor @theadhocracy @xvf17 @emilymbender You would REALLY benefit from reading "How to Win Friends and Influence People", because you can be right and still set people deeply against you, and it isn't their bias- it's very much your approach and attitude. A marketing person should already know this though.
People today can get good information from folks who aren't treating them with contempt. Why on earth would they talk to someone rude?
@SomeVeganCheeseIsOk @theadhocracy @emilymbender @Noisecolor
| Im not trying to win friends here.
That much is crystal clear.
Itβs also apparent that you donβt understand LLMs all that well.
@Noisecolor @xvf17 @theadhocracy @emilymbender How on earth do you think that makes you special here with us? Do you really think you're taking to the uneducated masses? Iguessed I had more experience in this field than you from your tone, and I'm not surprised to find that to be true.
You need to pull your head out of your ass.
@SomeVeganCheeseIsOk
Yes. It actually does seem like uneducated masses more or less.
That guy for sure. I doubt you as well.
@Noisecolor @xvf17 @SomeVeganCheeseIsOk @emilymbender No it's goalposts shifting, you're just changing the meaning of words π
And I'm not even sure why? Why do you **want** LLMs to be "AI"? If we actually get true AI, it'll be so much more than a statistical averaging model. Shouldn't we reserve the term for when it's applicable, and use terms relevant to the tools we have (like LLM)?
@theadhocracy
I'm making a point how the anti ai mob is silly. Nothing else. It's a pointless debate.
@Noisecolor @theadhocracy @xvf17 @emilymbender The problem being, you are being an asshole throughout this conversation, and that means nobody will listen or learn anything from you because nobody listens to assholes. So whatever your intent is here, you are failing at it because your approach sets people against you.
I was serious about you reading that book. If you want to be more effective, you need to be far, far less obnoxious.
@SomeVeganCheeseIsOk
You guys are the "it's not real pizza" or "rock is not real music" or "film is not real art" people.
That's how you look from outside when you say what you are saying.
And nobody cares if some people don't like rock music. That's fine. But trying to rationalize your own feelings like that is cringy.
I hope you get some time to think about it.
Let's end it, agree?
Prof. Emily M. Bender(she/her)
πππ ππππ β―οΈ β―
SomeVeganCheeseIsOk
Noisecolor
theAdhocracy