Discussion
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.
🧵>>
@emilymbender
Can i ask why you write <<"systems sold as "AI">> and then repeatedly you use that term again and again while you are referring to only one specific use of one specific technology under that umbrella? Maybe because you don't want to be "going against the mainstream"? Or maybe because having a childish simplistic stand against millions of different uses of thousands of different technologies is much more simple and comfortable than having to evaluate every one of them?
Must i refuse (amongst thousands of other things) to check the weather forecast because you think that by doing it i'm "carrying water for these clearly bad actors"? Don't you think your line of argumentation is wayyyyyyyyyy down the slippery slope?
@paelnever I say "systems sold as 'AI'" because I refuse to talk as if I thought "AI" was a thing.
I'm guessing you are not familiar with my work, but here's a good starting point:
@emilymbender
Sorry, maybe you are not familiarized with understanding what you read. I agree that "AI" isn't a thing, but in your toots you use that term again and again while you are talking of 1 use of 1 technology under that term.
I get that ranting against the whole "AI" is the trend but then don't start saying you think isn't a thing.
@emilymbender @wiert Oh. That's double jeopardy. Using it is harmful unless you can harm someone else responsibly. As if people are able to do so in the first place.
@emilymbender I think this might just be a coping strategy.
I used to just think this is garbage made in sweatshops through theft. I still mainly think that, but now what use is practically mandatory at work also sometimes I find myself looking up ”ethically sourced models”, ”local hosting”, and other such flights of hopefully fantasy.
My quests for such midde ground options have so far not amounted to anything useful, but it helps me feel slightly less complicit.
@emilymbender A lot of confusion could be avoided if people were specific about technologies and not dumping everything as AI and meaning LLM in the process. Lots of the success stories are not LLM, but ML algorithms or even statistical approaches that do not need an enormous investment, huge data centers and violation of copyright to work.
@emilymbender AI is basically irresponsible - especially because it was built by stealing all our data. Now they trying to make us pay for stolen goods. No way!
@emilymbender you might be interested in this article i just saw—it basically says we’re already doomed by the stupidity of a government that simply doesn’t care, as well as doesn’t really understand the problem:
https://www.blueamp.co/p/anthropic-fable-mythos-ban
@emilymbender I occupy that space, it is uncomfortable because it is not a popular stance, but I don't know that it is untenable.
I, based on my experience using these tools, believe they have strong potential to benefit humanity.
I am also aware of their issues, which I see as due to evil people holding the reins of society at the moment, not intrinsic to the technology itself.
@emilymbender This seems related to the phenomenon where many people feel the need to characterize themselves as on the right side of history (mostly to themselves I think) and so, working backwards, feel the need to explain how everything they do fits that idea. It is perhaps more productive to acknowledge that to varying degrees some of our decisions don't align well with our preferred values, but I don't know how to convince anyone of that.
a thing I've noticed recently is that no one here acknowleges that AI is helping cancer patients get better
@emilymbender
It's here.
It's not going away.
People will use it more and more.
It will not be meaningfully regulated, though the EU might attempt to.
@emilymbender do appreciate the thread, now i found it, and also still appreciate the ability to ask AI a question about a tech project that provides a mostly helpful, productive answer in minutes, that i’d never find (based on my prior tech experiences) in months of “reading the manual” (& which one?) or weeks of trying to get answers from folks who act snittty & entitled simply because they know more… i’ll still oppose energy projects utility payers have to subsidize since AI providers demand cheap prices to help offset an insufficient customer base, and i won’t pay for service levels i’ll never need or use. i am also expecting the problem to disappear as companies go belly up do to continual uncontrollable losses…
@emilymbender The argument is compelling. Now where are the "yeah but" brigade? I fancy entering the fray but I might need some training, and where better than here.
Come on you cowards, tell us how "inevitable" it is, give us your best shot.
@emilymbender Thank you so much for this thread Emily 💚 It's so good, I'm considering printing a paper copy to keep in my pockets for the next time I inevitably encounter theses arguments in the wild again.
Please let us know if you decide later to store this thread in an article form online. I would share this link with everyone!
@emilymbender Thank you so much for this thread Emily 💚 It's so good, I'm considering printing a paper copy to keep in my pockets for the next time I inevitably encounter these arguments in the wild again.
Please let us know if you decide later to store this thread in an article form online. I would share this link with everyone!
@emilymbender why is that untenable? i can see ecological or social reason to choose otherwise, but it seems rational to use tools when helpful even if they are dangerous when abused or otherwise misused (e.g. battery chainsaws, electric cars, etc.)
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.
>>
2- Righteousness. People do have legitimate needs, often unmet needs, and the synthetic text extruding machines can *look like* a solution. But just because the problems are real doesn't mean the solution is beneficial, effective, or worth (not always externalized) costs. Unfortunately, pointing out any of this is taken as the same as saying you don't care about the legitimate needs.
>>
@emilymbender This is actually the strongest one. An ESL job seeker in America is going to absolutely get creamed in the market, especially right now, when their imperfect English marks them as a no-hire. It's perfectly reasonable from their perspective to run their C.V. and cover letters through an LLMbecile to paper over the spelling, grammar, and improper idiom use cases.
Of course the real solution is for hiring managers not to be judgmental pricks, but that doesn't help the immediate need.
@emilymbender this was key in helping talk to a friend about their LLM use. Acknowledging the need and lack of social support and offering real human connection and aid instead helps them feel emotionally safe and heard. Then they're open to listening to the moral theme, because their needs have been acknowledged or met.
3- Whataboutism. This is used to brush off concerns about the externtalities of these systems. You eat meat, you fly on airplanes, etc, etc, how dare you talk about the impacts of data centers?
>>
@emilymbender
Makes sense to me for 2 reasons:
(a) if, in the 1920s, if a few folks had truly understood the Geography of Nowhere & environmental impact of widespread automobile usage, they'd have sought to prevent it by any means necessary (regardless of other bad acts going on).
(b) There is so little hard data (even projections) available that shows us how bad it's going to get vs. past environmental disasters.
I really wish we had (b) as settling the Whataboutism argument would be trivial.
@emilymbender I like to call the people who do this gotcha guys after this Matt Bors comic https://thenib.com/mister-gotcha/
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.
>>
5- Wishcasting. Some folks will point to scientific results from fields outside their own (usually media coverage thereof) that are marketed as having been done with "AI" and ask: How could you take a hardline against "AI" when it has provided XYZ?
>>
6- Exceptionalism. "I know this can be dangerous for people in general, but I know how to use it carefully."/"I know how to verify every output, and I am not deskilling myself." How do you know? Also, if you acknowledge the dangers to others, what example are you setting by talking about/talking up your use?
>>
So what is the best way out of that uncomfortable, untenable space? I think one key step is disaggregating the (non-coherent) set of technologies sold as "AI". If you don't call the stuff you work with "AI", you aren't saddled with trying to defend any of the rest of it.
>>
The most recent iteration of this conversation I was involved in turned in part on a strange, over-expansive definition of "genAI" which included, for ex, optical character recognition (OCR).
>>
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
The frustrating thing for me is how much of "old fashioned" machine learning gets lumped in to "AI" conversations these days, and then its (very real) usefulness gets used to prove the value of "AI," which is then used as an umbrella defense for setting LLMs on bullshit extrusion mode.
Even language models—possibly even large ones!—are very useful for translation, transcription, captioning, first passes of text annotation. i.e., jobs that involve interpreting language and language and doing language things with them.***
As you say, it's the generative nature of them—asking them to create or respond with meaning that wasn't in the input text itself—that starts the spiral out into resource-sucking noise.
(***maybe this sounds like the "middle ground" you're decrying but I think drawing a hard line at novel generation cuts off most of the hype nonsense.)
Install bonfire.cafe
Get the full app experience
