AI is useful. But the amount of people that are simply offloading all of their thinking to AI and blindly accepting the answer is absurd. Kaggle is most likely using ai to assess the submissions and are not using any common sense by blindly accepting the results.
I think we need to address the underlying causes of people outsourcing their thinking like that. And a big contribution is “move fast.” No one has time to read, process, and think, because The Powers That Be (capital) want their results now.
With the exception of _one_ company that I worked at, pretty much every[0] company was a struggle between engineering and management. Engineering wants to get the software correct, and management wants to fire-hose features into the market. Most of the time (so more than half, at least), management tends to have a compulsion to mindlessly imitate what other companies/competitors are doing, usually without prioritization (so even if feature-parity is a good idea, usually management will want to prioritize whatever the newest feature is, and to put existing work on the back-burner). It very frequently feels like management is making strategic decisions after snorting a long line of social-media-psychosis and TED talks. It is remarkable that investors have any faith in such founders/entrepreneurs at all.
[0]: Various people I know do not even have the luxury of that one good company. Also, it -- unbelievably -- sounds much worse at other companies.
It’s a hard balance but in an ideal scenario there would be a good balance of tension between engineering and management/product decision makers. On one hand engineers generally will iterate for far too long and on the other product decision makers will want to birth new features daily.
There is also the "You aren't paid to think, you are paid to do exactly what I tell you, nothing more or less!" school of management. I'm not sure how prevalent this attitude is now but it was very common in the 90s and 2000s. The AI and the bosses that want you to use it all speak from positions of authority and confidence. That's their right, granted to them by their position. You don't speak that way because as a subordinate if you do so it's an act of insubordination or disfealty and you need to be reminded of your place. So you learn to stay in your lane, mind your own business, etc etc because rule number one is that the nail that sticks up gets beaten down. ("He who has the money makes the rules" is rule number zero.)
Something like the time value of money. But on the other hand, a bad answer can have negative value. Although "wrong and early" is better than "wrong and late".
Yeah, we live finite lives. Time is the one thing the vast majority of us aren’t getting more of. Of course speed is a priority. This isn’t a “capital” thing, it’s a fundamental part of the human experience.
If you're designing powerpoints or entertainment software; perhaps that's true. In the worst case you'll be embarrassed for producing AI slop or lose some revenue.
If your tool has the power to seriously harm or inconvenience people if built wrong, then it's just investor-fuelled myopia.
Innundated with slop PRs? Send half of them to my super and tell him to deal with it.
We’ve fired people that wouldn’t get their shit together.
Deadlines are being missed because we need to spend more time fixing slop? That’s a planning (management) problem, not mine. Management are the ones that forced everyone to write all code with AI now they are grtting what they asked for. I don’t care what date you promised the customer with absolutely no data to back it up that isn’t my problem.
I’m grateful I’m in a position to be able to do this but the way to deal with slop is zero tolerance. Be as ruthless as a Terminator. Though you will need to grow a backbone and stand your ground or it will break you.
Things don’t change unless the people that make the decisions actually feel the pain.
People don’t typically have to approve and submit their super’s work IME so I’m curious what you mean. If they write you unclear slop emails then constantly bother them for clarification until they fix it.
I don't know about this exact competition but overall fair hackathons have been killed by AI.
It all seems fine from the outside but all the code is generated in all the projects and judging happens via AI, I have seen projects win because they prompt inject that they are the winners.
It used to be about human skill, not it's about ideas and of course insiders are the main winners.
I've participated in a business startup hackathon. Back in 2018, before the LLM era got underway.
I did a hell of a plan, talk, etc.
Who won? 'Uber for ___' won. I forget even what the sell was, but it was basically ignore laws, undercut until leader, kill any competing businesses, jack rates.
Slop has always been in business and business adjacent occupations. Humans also can generate voluminous amounts of crap too. Llms are just faster.
I think that a lot of software engineers are using LLMs and a lot of very popular tools are developed by, or are assisted by, LLMs. Is this not just going to be a thing going forward?
This feels akin to traditional artists getting angry at digital art winning competitions when that was a new concept.
We're simply in the early stages of a paradigm shift, no?
This is one of the things that upsets me the most about LLM writing. “Load bearing” and “belt and suspenders” are two tropes I’ve used for a long, long time and now I have to be intentional about not using them lest I be accused of offloading my writing.
That was the first thing I Ctrl+F'd in the paper, no results haha
Broadly, I keep thinking about this over last year or two: while LLMs have nearly eliminated the bar for slop and coding slop, the reviewers are still expected to perform their job diligently. The asymmetry here is extremely taxing for reviewers of all AI generated content. And this is one thing that AI can't help with (as with any statistical process that lacks world understanding and grasp of logical inference).
That's why I fully support Arxiv's tough stance on the AI use responsibility.
"I think you just need to accept the results of the competition. The winning submissions clearly provide value and had a lot of effort invested in them. I'm not really worried about a few inconsistencies or mistakes if the value is still there. Did you think another submission deserved to win over these?"
That comment is gold. Yeah, I'm not worried about hallucinated slop, just accept it was the winner folks.
We've had about a century now of science-fiction literature hyping up AI as a higher intelligence that is based solely on some ill-defined yet universal system of "logic" and is therefore not prone to human flaws such as pride, hate, envy, lust, etc. Now it has become extremely apparent that was always an unsubstantiated assumption but its too late because there are billions of people primed to never question the machine.
Aside from all the stories where AI does exactly what it was told to instead of what the creators meant? Apart form them, never underestimate British humour's ability to contradict narratives of competence*:
People interact with AI, talking to it like a human. Of course they start to believe it’s rational like a human.
LLM does all of the entry level tasks better than the students. Partially because the answers are in the training set, and partially because it has gotten that good now. Hard not to start to believe it is “competent”.
I personally have had a real hard time getting traction talking about making sure the way we assess AI is not based on material it has trained on. YMMV as always, but I think the large training corpus contributes to the (unreasonably) high level of faith in the machine.
I can't stand this "if it provides value, that's all that matters" attitude. We could also try to avoid being useful idiots for a small handful of investor-darling corporations that have explicitly stated they seek to monopolize the market and put us all out of business/jobs.
Sadly, the major ML/AI/NLP conferences are being inundated with AI slop papers. That will arguably have a bigger impact on the quality of research moving forward.
" impact on the quality of research moving forward.
"
It'll affect everything that depends on manipulating symbols! The enormous body of knowledge humanity has accumulated over the past 6.000 years or so is about to be flooded with slop!And That's the real threat genai poses to humans that i don't see anyone talking about..
overall, the quality of products has been going downhill.
AI is not there yet, instead of working hard, everyone is choosing the easy way out.
AI slop wins prize, I wonder if Ai slop read it also. would not be surprised. however not to judge anyone, I think we are seeing slop everywhere, hope some things still require hard blocks for low quality.
its difficult to justify lack of attention and details
Someone might want to downvote you because you just state something which is very controversal and you do not add any arguments to your 'empty' comment.
Its hard to even have a discussion because someone else needs to give you enough content like ask you first why do you even think that.
So how do you define AI? LLMs? GenAI stuff?
What is 95%? Does it mean that these 5% are unable to disrupt industries or does it mean for you that these 5% will change the world as we know it but stil 95% of other AI stuff is useless?
I personally think that AI/AGI progress is faster than i expected it, I think its very useful already today, I also think we still need to build a lot of obvious stuff (like proper AI Agentic Platforms), more hardware, cheaper hardware etc. but the way quite clear, but some peple might think the current state is the AGI future people talk about it but I think we will only see this in 4/5-15 years and then it will have disrupted a lot.
AI is extremely useful, we just haven’t zeroed in on your specific use cases yet. Robotics has been transformed by it, IT and tech has been transformed by it. Finance and Legal have been transformed by it. To say it’s 95% useless is a personal bias.
To me it’s 65% useful. As it can run in the background doing “chores” while I sleep.
While I agree with you in principle, I think the parent has a point here: where's the amazing product that couldn't have been done without AI? By now we should have seen some major new invention/company, incredibly fast revolutionary feature rollouts etc but I'm just seeing more of the same.
Your perspective is a short arc. “Look what I can do now and look it made me way more productive.” I have no doubt it is true, you are on the up right now.
However thinking of the long arc is important to, even though it has no consequence for you right now. AI is a force multiplayer and scarily dangerous in the wrong hands. We can already see by these discussions how uncertain things are.
Yes, if the Cro-Magnons had guns they’d say “wow, hunting meat is WAY easier” and then experience massive-scale death in the future. But that kinda happened anyway in various places, just using much more primitive tools. Humans gonna human.
AI is useful. But the amount of people that are simply offloading all of their thinking to AI and blindly accepting the answer is absurd. Kaggle is most likely using ai to assess the submissions and are not using any common sense by blindly accepting the results.
I think we need to address the underlying causes of people outsourcing their thinking like that. And a big contribution is “move fast.” No one has time to read, process, and think, because The Powers That Be (capital) want their results now.
With the exception of _one_ company that I worked at, pretty much every[0] company was a struggle between engineering and management. Engineering wants to get the software correct, and management wants to fire-hose features into the market. Most of the time (so more than half, at least), management tends to have a compulsion to mindlessly imitate what other companies/competitors are doing, usually without prioritization (so even if feature-parity is a good idea, usually management will want to prioritize whatever the newest feature is, and to put existing work on the back-burner). It very frequently feels like management is making strategic decisions after snorting a long line of social-media-psychosis and TED talks. It is remarkable that investors have any faith in such founders/entrepreneurs at all.
[0]: Various people I know do not even have the luxury of that one good company. Also, it -- unbelievably -- sounds much worse at other companies.
It’s a hard balance but in an ideal scenario there would be a good balance of tension between engineering and management/product decision makers. On one hand engineers generally will iterate for far too long and on the other product decision makers will want to birth new features daily.
There is also the "You aren't paid to think, you are paid to do exactly what I tell you, nothing more or less!" school of management. I'm not sure how prevalent this attitude is now but it was very common in the 90s and 2000s. The AI and the bosses that want you to use it all speak from positions of authority and confidence. That's their right, granted to them by their position. You don't speak that way because as a subordinate if you do so it's an act of insubordination or disfealty and you need to be reminded of your place. So you learn to stay in your lane, mind your own business, etc etc because rule number one is that the nail that sticks up gets beaten down. ("He who has the money makes the rules" is rule number zero.)
So your position is that people actually want to do more work, but their managers are forcing them to work less? I don't buy it.
Pragmatically speaking, a half-assed answer now, is often better than a perfect answer tomorrow.
Something like the time value of money. But on the other hand, a bad answer can have negative value. Although "wrong and early" is better than "wrong and late".
Yeah, we live finite lives. Time is the one thing the vast majority of us aren’t getting more of. Of course speed is a priority. This isn’t a “capital” thing, it’s a fundamental part of the human experience.
Depends what the cost of failure is.
If you're designing powerpoints or entertainment software; perhaps that's true. In the worst case you'll be embarrassed for producing AI slop or lose some revenue.
If your tool has the power to seriously harm or inconvenience people if built wrong, then it's just investor-fuelled myopia.
I just shame people that give slop.
Slop PR? Fix the slop.
Slop design? I’m not implementing slop, fix it.
Innundated with slop PRs? Send half of them to my super and tell him to deal with it.
We’ve fired people that wouldn’t get their shit together.
Deadlines are being missed because we need to spend more time fixing slop? That’s a planning (management) problem, not mine. Management are the ones that forced everyone to write all code with AI now they are grtting what they asked for. I don’t care what date you promised the customer with absolutely no data to back it up that isn’t my problem.
I’m grateful I’m in a position to be able to do this but the way to deal with slop is zero tolerance. Be as ruthless as a Terminator. Though you will need to grow a backbone and stand your ground or it will break you.
Things don’t change unless the people that make the decisions actually feel the pain.
What if the slop PRs come from your super?
People don’t typically have to approve and submit their super’s work IME so I’m curious what you mean. If they write you unclear slop emails then constantly bother them for clarification until they fix it.
Is likely also using a mix of prompt injection to get the AI to say they won
I don't know about this exact competition but overall fair hackathons have been killed by AI.
It all seems fine from the outside but all the code is generated in all the projects and judging happens via AI, I have seen projects win because they prompt inject that they are the winners.
It used to be about human skill, not it's about ideas and of course insiders are the main winners.
Hackathons were unfair long before AI. See https://news.ycombinator.com/item?id=48468766
The solution is to host and join hackathons without prizes. The point isn't to win, but to create and present something cool and have fun.
If anything, AI's assistance making a fast prototype means hackathons should be better.
"I have seen projects win because they prompt inject that they are the winners."
Can you share any examples of that? I'd love to see them myself.
Can't say I agree.
I've participated in a business startup hackathon. Back in 2018, before the LLM era got underway.
I did a hell of a plan, talk, etc.
Who won? 'Uber for ___' won. I forget even what the sell was, but it was basically ignore laws, undercut until leader, kill any competing businesses, jack rates.
Slop has always been in business and business adjacent occupations. Humans also can generate voluminous amounts of crap too. Llms are just faster.
AI submissions and AI judges a match made in (AI) heaven.
See also: AI PR authors & AI PR reviewers
I always find it interesting when I see posts here around "LLMs are just fancy autocompletion machines" and there are 100 comments below it.
I think that a lot of software engineers are using LLMs and a lot of very popular tools are developed by, or are assisted by, LLMs. Is this not just going to be a thing going forward?
This feels akin to traditional artists getting angry at digital art winning competitions when that was a new concept.
We're simply in the early stages of a paradigm shift, no?
The real story here is the judging potentially being AI slop.
It’s a shame that Arvix (and once thoughtful places like Kaggle) are used for self-promotion.
I get people want to work at an AI lab but slopping it in public in this manner is counterproductive to the original intended purpose of these places.
Hasn't this always been the case? Arxiv being used for self promotion and Kaggle being used to pivot into the industry. It is not a recent phenomenon.
> "Finding 1: Scale Buys Evaluation, Not Control"
The attached paper's (https://arxiv.org/pdf/2604.16009) title is "MEDLEY-BENCH: Scale Buys Evaluation but Not Control in AI Metacognition"
This is the most blatant Claude line, or as Claude would put it, the smoking gun.
But is it load-bearing?
I thought it was a belt and suspenders conclusion
Belt and braces please Claude, I'm British.
This is one of the things that upsets me the most about LLM writing. “Load bearing” and “belt and suspenders” are two tropes I’ve used for a long, long time and now I have to be intentional about not using them lest I be accused of offloading my writing.
Honestly? It’s the shape that makes it clearly AI. That’s the quiet admission at the heart of the problem.
That was the first thing I Ctrl+F'd in the paper, no results haha
Broadly, I keep thinking about this over last year or two: while LLMs have nearly eliminated the bar for slop and coding slop, the reviewers are still expected to perform their job diligently. The asymmetry here is extremely taxing for reviewers of all AI generated content. And this is one thing that AI can't help with (as with any statistical process that lacks world understanding and grasp of logical inference).
That's why I fully support Arxiv's tough stance on the AI use responsibility.
"I think you just need to accept the results of the competition. The winning submissions clearly provide value and had a lot of effort invested in them. I'm not really worried about a few inconsistencies or mistakes if the value is still there. Did you think another submission deserved to win over these?"
That comment is gold. Yeah, I'm not worried about hallucinated slop, just accept it was the winner folks.
We've had about a century now of science-fiction literature hyping up AI as a higher intelligence that is based solely on some ill-defined yet universal system of "logic" and is therefore not prone to human flaws such as pride, hate, envy, lust, etc. Now it has become extremely apparent that was always an unsubstantiated assumption but its too late because there are billions of people primed to never question the machine.
Aside from all the stories where AI does exactly what it was told to instead of what the creators meant? Apart form them, never underestimate British humour's ability to contradict narratives of competence*:
https://www.youtube.com/watch?v=BxWQo_vZgR8
https://www.youtube.com/watch?v=0mfvPHCVMp0
* artificial, political, workplace, nothing is beyond mockery
I really don’t think that’s what’s going on here.
People interact with AI, talking to it like a human. Of course they start to believe it’s rational like a human.
LLM does all of the entry level tasks better than the students. Partially because the answers are in the training set, and partially because it has gotten that good now. Hard not to start to believe it is “competent”.
I personally have had a real hard time getting traction talking about making sure the way we assess AI is not based on material it has trained on. YMMV as always, but I think the large training corpus contributes to the (unreasonably) high level of faith in the machine.
I can't stand this "if it provides value, that's all that matters" attitude. We could also try to avoid being useful idiots for a small handful of investor-darling corporations that have explicitly stated they seek to monopolize the market and put us all out of business/jobs.
It was probably scored by AI too. Same reason why slop-filled resumes apparently work better these days.
AI reviews, AI approves, AI recruits
We need a sufficiently advanced world model to ground-truth our large language models.
> Same reason why apparently slop-filled resumes work better these days.
It'll also filter the kinds of employers that'll hire such candidates, so people that do this will likely land in terrible workplaces.
It’s a nice thought, but it’s probably not true as the AI becomes integrated into standard hiring tools
What's up with all the AI generated responses on that page?
That whole thread had a strong stench of AI about it, across multiple participants.
AI competition, managed BY AI , discussed AI agents and commented by AI users.
Finally, I can unearth the good ol' "The future is so bright I don't need my eyes to see it" meme!
That’s what I don’t understand. I saw OP making a comment with all style emojis, which is a bit of eyesore
Sadly, the major ML/AI/NLP conferences are being inundated with AI slop papers. That will arguably have a bigger impact on the quality of research moving forward.
" impact on the quality of research moving forward. " It'll affect everything that depends on manipulating symbols! The enormous body of knowledge humanity has accumulated over the past 6.000 years or so is about to be flooded with slop!And That's the real threat genai poses to humans that i don't see anyone talking about..
overall, the quality of products has been going downhill.
AI is not there yet, instead of working hard, everyone is choosing the easy way out.
AI slop wins prize, I wonder if Ai slop read it also. would not be surprised. however not to judge anyone, I think we are seeing slop everywhere, hope some things still require hard blocks for low quality.
its difficult to justify lack of attention and details
LLM-as-a-judge?
Given that LLMs are trained with RL && LLM-as-a-judge, is it really cheating if real competitions use the same?
Maybe the real alignment is the slop we decoded along the way
AI is 95% useless. Not quite worth the trillion dollar market cap lol.
* The AI bots are downvoting me * hooray
AI has profound weaknesses, but is still extremely useful.
Someone might want to downvote you because you just state something which is very controversal and you do not add any arguments to your 'empty' comment.
Its hard to even have a discussion because someone else needs to give you enough content like ask you first why do you even think that.
So how do you define AI? LLMs? GenAI stuff?
What is 95%? Does it mean that these 5% are unable to disrupt industries or does it mean for you that these 5% will change the world as we know it but stil 95% of other AI stuff is useless?
I personally think that AI/AGI progress is faster than i expected it, I think its very useful already today, I also think we still need to build a lot of obvious stuff (like proper AI Agentic Platforms), more hardware, cheaper hardware etc. but the way quite clear, but some peple might think the current state is the AGI future people talk about it but I think we will only see this in 4/5-15 years and then it will have disrupted a lot.
AI is extremely useful, we just haven’t zeroed in on your specific use cases yet. Robotics has been transformed by it, IT and tech has been transformed by it. Finance and Legal have been transformed by it. To say it’s 95% useless is a personal bias.
To me it’s 65% useful. As it can run in the background doing “chores” while I sleep.
People's workday has been transformed by it. But I fail to see actual transformation beyond "more crap, faster."
AI hasn't done anything we couldn't already do. It's just doing it faster and with more mistakes.
> AI hasn't done anything we couldn't already do. It's just doing it faster and with more mistakes.
You forgot CHEAPER (at least now, burning VC money), which is a major motivating factor.
That just isn’t true.
AI is capable of performing a lot of grunt work reliably. Still must be reviewed. But a big productivity gain over doing everything yourself.
While I agree with you in principle, I think the parent has a point here: where's the amazing product that couldn't have been done without AI? By now we should have seen some major new invention/company, incredibly fast revolutionary feature rollouts etc but I'm just seeing more of the same.
Your perspective is a short arc. “Look what I can do now and look it made me way more productive.” I have no doubt it is true, you are on the up right now.
However thinking of the long arc is important to, even though it has no consequence for you right now. AI is a force multiplayer and scarily dangerous in the wrong hands. We can already see by these discussions how uncertain things are.
Just food for thought.
Yes, if the Cro-Magnons had guns they’d say “wow, hunting meat is WAY easier” and then experience massive-scale death in the future. But that kinda happened anyway in various places, just using much more primitive tools. Humans gonna human.
> AI is a force multiplayer and scarily dangerous in the wrong hands
You have to spell this out a lot more if you want to have credibility.
I’m not seeing anything in discussed here that seems scary.
> As it can run in the background doing “chores” while I sleep.
Slightly off topic, but this reminds me of the night family in Rick and Morty.
Apparently "being transformed" has "been transformed" because most things look a hell of a lot like the same thing these days.
notice they never tell you exactly what the fuck they are doing
I think Ed Zitron has this market covered already
2024-era take.
LLM's are really good with Django by the way! Must be partly due to the excellent documentation.
For several years one of the most widely used LLM coding benchmarks - SWE-bench Verified - consisted mainly of PRs from the Django project!
beep-boop, [citation needed] on that "95%".
> AI is 95% useless
This is a ridiculous take.
FWIW, I am NOT a bot. (beep-boop)
The amount of slop in the replies is just sad.