I attended a conference recently and AI use by academics was absurd
81 points by oceanhaiyang
81 points by oceanhaiyang
As we listened to talks most people were stuck doing pointless tasks on their phones and computers. Less than 10%, likely 5%, were listening attentively. This was quite sad for me.
The thing I would like to share the most, that shocked me more than any AI use I've seen ever—when I was on stage receiving QA post talk among the fellow panelists, everyone but me used AI on their laptops to record their questions and then come up with a response they could use. IDK about others, but I spent a lot of time on my talk and slideshow. The professor who served as the respondent listened for two hours taking a ton of notes then formed thoughtful questions. I cannot believe the laziness and disrespect to the professor's intellect for all the other panelists to rely on AI for their response.
For me, the seduction of AI to do boring tasks such as translation, CV writing, and job applications, and editing papers is forever there but I choose to believe these tasks benefit me as someone trying to be a so-called intellectual. To see everyone around me choose to not think for themselves, when we presenters should be the most knowledgeable in the room about our topics, shocked me beyond anything I've seen in academic.
I am constantly considering the use of AI in the classroom and research and thought this reflection/rant would be relevant here.
I'm curious if my fellow Lobsters have seen similar uses of AI or what their opinions are.
Once LLMs are being used to communicate in place of a human response, I want out. It happened in my last job from my direct manager, it's started happening in parts of my current job with my own peers. I also don't understand it, it's removing the humanity from our society.
Fully agree. I've never witnessed people using it beyond cheating on exams and essays. This is really dark...
The same day this happened I applied for a job including a proposal against tech usage in the classroom, of course including AI, to support the molding students' own critical thinking.
Yeah, even as a person who loves LLMs, I gotta agree with that.
If I'm talking with a person, it's because I'm actually interested in what they think. It's pretty rude when they turn out to be a proxy for their AIs. There is no point in the social interaction if they're just middle men. I can just chat with the AIs myself.
Human interaction is not synonymous with effort and good faith though. Sometimes people put approximately zero effort into engaging with what you said, and that's just as disrespectful as an AI response. I'd say bringing out the so called clanker is a perfectly valid response to that energy. Better than wasting words on people who'd just ignore them.
There's an issue with speed and expected value though.
Clankers can get a pretty good response out, virtually instantly. Let's say 90% coverage (of a high effort, introspecting consideration by an intelligent human) in 1% of the time?
I feel like we should be talking to clankers more for trivialities (i.e. a discussion with a coworker on the rationality of buying insurance and Jensen's inequality). With the downfall of Google (and by proxy, DDG and other search engines), local LLMs + various providers have become my substitute for "how to X". Maybe we need a new lmgtfy, let me clanker that for you?
My old rule was always "make sure any question you ask cannot be answered by a 5 second google search" and I feel like that respect and consideration should carry on into the LLM era. Perhaps some queries deserve clanker responses?
I really think you nailed the whole thing with the word "disrespect." I have been thinking a lot about the human side of technical work - mainly the communication. I've had an experience, well two experiences, with the same person sending me a big copy and paste of LLM output to me. The first time I read through all the points, took time to craft responses, and hopefully foster an actual discussion. It went nowhere. The second time they did this I simply told them I will not read it and we can get on a call to talk about the points if they want. They did not want. Who benefits from either of these situations? No one.
To waste a person's time with having them read, parse, and interpolate ideas that you, yourself, have not even come up with is extremely disrespectful.
High-school teachers are very similar.
I was once in a staffroom where the teacher raved about using AI to mark assignments for him during the morning tea break, and was furious that his students were using AI to write their assignments at lunch break.
He didn't see the irony, and pointing it out was not very well received.
Tired: We pretend to work, and they pretend to pay us.
Wired: They pretend to learn, and we pretend to teach them.
Why even go to a conference if not to talk with people directly? Like, what on earth do you get out of it.
If it is an academic conference, it may be an obligatory part of the process: your productivity is measured by the papers you publish, and in CS that is usually at conferences (as opposed to journals), which means that you present your paper at that conference. You might make use of the opportunity to listen to other presentations and talk with people (and it may even be good for your career), but that's a choice.
Though, this doesn't quite fit with OP's use of "panelists". So I guess this situation is different.
when I was on stage receiving QA post talk among the fellow panelists, everyone but me used AI on their laptops to record their questions and then come up with a response they could use.
Can you elaborate on this part? Like attendees would ask a question, and the panelists would just repeat the question to an AI chatbot and read out what AI said?
I assumed this was what the sentence meant too. This was startling unless the speakers were non-native English speakers and were very concerned about getting the phrasing correct.
Part of the value add of a face to face conference is fluid and spontaneous discourse where the people's personalities come out.
I am aware of a trend at least among undergraduates where non-native speakers will come to class, set up a phone to record continuously, and watch live translations. Obviously this is problematic in its own ways but it's at least understandable—I wonder if this was the baseline that made this not seem abundantly weird but then they took it a step further to have it automatically formulate responses?
The LLMs are training "us" now.
First we develop the machines, then we contort the entire social and psychic order to serve their rhythms and facilitate their operation.
I think we are going to be seeing a vast partitioning in society in the next months and years.
The process of forming expressions just is the process of conceptual and rational articulation (as per Brandom). Those who misunderstand this -- believing that concepts are ready made, then encoded and decoded from permutations of tokens, or, worse, who have no room to think of reasoning or conceptualization at all -- they will be automated away.
I don't mean that their jobs will be automated: I mean that they will cede sapience and resign to becoming robotic. A robot is just a "person whose work or activities are entirely mechanical" (https://www.etymonline.com/search?q=robot).
I'm afraid far too many are captive to the ideology of productionism (which is just a corollary of consumerism). Creative activity is not about content production. The aim of our creation is communication and mutual-transformation. Generation of digital artifacts may be useful for these purposes, but most uses seem to assume content production is the point, and that is a dark, sad, dead end.
I think LLMs are an extreme force modifier on the pitfalls and dangers of technology as elucidated by the likes of Heidegger, Ellul, Baudrillard, and Kacynski. On a society-wide level their explosion has had wide effects that disconnect us from each other, nature, and reality.
The extreme individual examples of this are people who lose touch with reality via AI psychosis or filter their every human interaction through an LLM, but more common is seeing people who used to care about their work become apathetic about quality and lower their own standards to that of slop. And that apathy can extend to having a toll on them personally.
Yes, I agree! LLMs just accelerate a process that has been underway for a very long time. I would had Mumford, Ursula Franklin, and Adorno to your list!
Sorry for the meta comment but this is perhaps the greatest example I've seen of why "ai" and "vibecoding" are insufficient as tags. This isn't about AI development, it's AI usage, so it's "vibecoding." But it doesn't even mention coding!
Yeah, I think this is a weird part of lobste.rs culture. Tags are used mostly for filtering out topics, and so tags are mostly used by people who don't like that topic. You can even see this in the UI for suggesting tags: the one piece of information you get is how many people are filtering each tag. As a consequence, the tag wording tends to be either derogatory ("vibecoding") or needlessly obtuse in an attempt to avoid hype language ("merkle-trees" instead of "blockchain"). I've just accepted it at this point, it's the law of the land I suppose.
tag wording tends to be either derogatory ("vibecoding") or ...
Without otherwise disagreeing with your comment, I note that that term was coined as promotion, not criticism.
True, but it was used in promotion of a specific practice, which is using AI for stuff you don't really care about, never looking at the code and generally full YOLOing it. Applying it to any use of AI for software development is therefore derogatory in my view. Not that that's necessarily bad, after all I'm a big AI hater myself as well and don't code with AI at all. I just to wanted to describe a weird state of affairs I see where such content is tolerated here but only on condition of being ridiculed as vibecoding.
What's derogatory is the definition Lobsters uses; the original definition of "vibecoding" was "intentionally using AI in a slapdash way," and the Lobsters tag is designed to disparage any use of AI in programming as innately slapdash no matter how much care went into it.
I find that AI as an appropriate tool like anything has its place. It shouldn't do the thinking for me, nor should it be my voice. In my own writing, I typically have my own section and then clearly indicate when I'm using the llm to document the more boring stuff. As far as disrespect goes, I think it goes hand in hand with cell phone use, lack of attention due to distraction, and social media. So I would see it more as a continuation of an old problem rather than a new problem in and of itself.
… this reflection/rant …
It's a reflection, in no way a rant (to my eyes). Thank you for expressing things fairly respectfully.
… To see everyone around me choose to not think for themselves, …
That has the potential to insult people who choose to think with their use of AI.
More importantly, it underestimates the degree to which appropriate use of AI can be beneficial.
Recovering academic, friends, family & former colleagues give me visibility on wide swathes of fields (biology/medicine, mathematics, computer science).
Besides the conscientious objectors, AI use is rampant. Particularly painful for grad students who are feeling mounting pressure to perform.
A sad reality I discussed with an nearing retirement-age professor (quite famous, relatively) was delegating simple proofs/lemmas to ChatGPT instead of a grad student. The instant gratification is just too tempting. He knew it would come at a training (students, not AI) cost, but when you're trying to get a paper out, the adjacent goals become somewhat blurry.
Academics are probably particularly susceptible to AI because of high pressure environments/monofocal type personalities. A friend's son who is a postdoc mentioned a lab of a quite famous (Nobel/Fields) tier scientist who will remain unnamed where students felt the need to constantly have prompts/agents running exploring the problems they're working out, lest someone "scoop" them and get the paper and recognition out first. For some less collaborative fields with no lab/data gathering requirements, I suspect it's really reducing incentives to work together/divide work (and credit!). Relevant publicized anecdote was the quantum cryptography results that two researchers "found" independently (with help of GPT 5.6 Sol) and thereafter decided to collaborate on.
Another thing that a lot of people have commented on is the desertion of office hours and reduction in collaboration among undergrads.
As we listened to talks most people were stuck doing pointless tasks on their phones and computers. Less than 10%, likely 5%, were listening attentively. This was quite sad for me.
It was already like this before LLMs, and not just in academia (eg. tech conferences or work meetings). I also found it disrespectful and make sure I either keep my laptop closed for most of the talk, or get out. I don't understand why people go to a talk not to listen to it.
when I was on stage receiving QA post talk among the fellow panelists, everyone but me used AI on their laptops to record their questions and then come up with a response they could use.
wow. I left academia before LLMs and didn't realize it was this bad. I guess the next step is this satirical post becomes reality?
I always assumed they were like me - I like to listen to talks and do some not so demanding work. One part of my brain listens and the other part does the other work. I stop when something really attention needing comes up.
It's kind of hard to know why some people are on their laptops. I've sat next to some really engaged note takers at conferences, but also some people just doing other work and you can't always tell them apart.
There's probably a bit of neurospicyness involved too in science/computer circles. Unless the talk is extremely engaging, I'm likely listening more and not getting distracted if I do some mindless thing in the background.
What type of conference was it? I've been to a couple of condensed matter physics ones recently and saw nothing of this, all the presenters were professional and knowledgeable without an AI crutch.
Less than 10%, likely 5%, were listening attentively.
To be honest... this is my experience with major academic conferences prior to LLM normalisation as well. It has certainly given everyone a specific kind of (un)interaction, though, and I've seen it invade all corners of our work. Given that the point of academia is to strive for personal and professional excellence in both product and practice, it just seems totally antithetical.
I don't subscribe to the idea of resisting AI to save the human side of things. I consider that to be a silly endeavour that is rooted in confusion.
If the outcome of using AI for a given task is good, then it is a benefit to whomever uses it. You wouldn't oppose people using industrial manufactured cookware because it was once hand made.
The question is and always will be if the outcome is better or not. Which I think it's something almost everyone is biased pro-AI. People are so charmed with what is possible all of the sudden, that they don't even stop tone evaluate the quality of the output.
My problem with AI generated works is not that they are generated using AI, is that by and large they suck big time.