top 100 comments

sorted by: hot top controversial new old
[–] 139 points 1 month ago* (last edited 1 month ago) (20 children)
  1. is an insanely fast learner

This person very clearly does not understand, even at a basic level, how llms work. They are completely incapable of "learning" outside of training, and training cannot happen while you are using them - in no small part because they require MILLIONS of tokens of data (minimum) to be able to recognize patterns.

Meanwhile that theoretical stupid kid she thinks should be paying her masters to have her teach them can learn instantaneously from even a single datapoint.

  • source
  • hideshow 20 child comments
  • [–] 63 points 1 month ago* (last edited 1 month ago)

    She's wrong about the advantages of AI, but she's right about how companies are going to see it. It's true that they'd rather deal with AI than an intern, and it's true then that internships will fall onto the universities to coordinate like rotations as part of an expensive education system.

    It's also true that kids are learning how to use AI on their own, and are losing their ability to learn from any datapoints, because why do that if the magic box will give you an answer that sounds correct enough? Ultimately, it doesn't matter how useless AI is if everyone acts as if it's as useful as this lady thinks it is.

  • source
  • parent
  • [–] 7 points 1 month ago (16 children)

    I'm told that agents these days keep a scratchpad of notes to themselves, so an agent can (kind of) learn about things by making a note about it which it will still have access to later, even outside the formal context window.

    It's going to be far more limited than the kind of training you're talking about, but it is something which an individual agent can do

  • source
  • parent
  • hideshow 16 child comments
  • [–] 32 points 1 month ago* (last edited 1 month ago) (14 children)

    That is sort of true, but it still isn't learning. Basically, it can store information usable by other sessions. But to do so, it just processes that information the same way as a prompt. So less like learning and more like just tacking something on to its context window. So, it never "learns" that information, it just tells itself that thing every time it processes any prompt.

    (don't downvote MartianSands' comment - it's a good point. Shame on you)

  • source
  • parent
  • hideshow 14 child comments
  • [–] 5 points 1 month ago (12 children)

    Explain the distinction in more detail, and why it matters.

  • source
  • parent
  • hideshow 12 child comments
  • [–] 31 points 1 month ago* (last edited 1 month ago) (9 children)

    Say you had a version of chat GPT that hadn't been trained on a corpus that included enough information about the number of Rs the word "strawberry" contains, didn't have workarounds for character counting, and has context sharing.

    You ask it "How many Rs does Strawberry contain?"

    It burns a bunch of tokens, returns "Strawberry has 7 Rs. No wait, that's not right, strawberry has 5 Rs. No wait..." etc.

    You say "Strawberry has 3 Rs. How many Rs does Strawberry have?"

    It says "Strawberry has 3 Rs."

    You ask "How many Rs does this exact string contain: 'Strawrberrrry'?"

    A new session using shared context responds "Strawberry has 3 Rs."

    You say "No, it has a different number of Rs. I didn't ask about Strawberry this time"

    It responds "Sorry about that! You didn't ask about Strawberry, you asked about Strawrberrrry. Strawrberrrry has 7 Rs. No wait, that's not right. Strawrberrrry has 5 Rs...."

    LLMs, at their core, use models that have computed lexical & semantic similarity to predict words (really they perform contextualized vector transformations, but let'snot get into that). When they "learn", they just do this more effectively using more text or using the same text more efficiently. Next word prediction is not the same thing as learning concepts, like the concept of numbers, non-numerical things being ascribed numerical values (ironically, since LLMs function by turning words into numerical vectors), or counting.

    When it responds "Strawberry has 3 Rs" it isn't because it knows what that means or how to gain that knowledge about other words. It is merely parroting back what you've told it because "Strawberry has 3 Rs" in its shared context has very close lexical similarity to your query "How many Rs does Strawberry have?". It also parses Strawberry as ["straw", "berry"] encoded into numerical values representing their relationship in the corpus' vector space (e.g. {[ .420, -.67, .67], [.420, .69, -.69]}) - so unless it has instructions to further break those tokens into characters, it does not have the capability to count. You can actually do some weird math using these vectors and their relationships (their distance apart in the 3D vector space, angles between vectors, etc actually relate to the semantic content of the tokens), but you lose granularity such as the number of letters in a token when you look at words this way.

    Learning is more than just computing lexical similarity. Modern LLMs mimic reasoning by expanding the query with intermediary tokens, basically creating a temporary scratch pad of related words on the same vector, but don't actually reason (this is known as chain of thought). Learning is partly computing lexical similarity, but it is also about extrapolating concepts from facts and inferences, applying those concepts to novel simuli, etc. LLMs really don't do reasoning, which is part of learning.

  • source
  • parent
  • hideshow 9 child comments
  • [–] 11 points 1 month ago (1 child)

    I would also like to add another key difference between human learning and machine learning. When I, as a person, read a book, I can conceptually understand the story. A machine would digest that same book by, for example, counting the numbers of times each word comes after another and assigning probabilities to occurrences or words or basic sentences. The major differences are still being explored. We don’t fully understand how humans learn, fundamentally. We know somewhat mechanisticly that neurons share and process chemical signals, but we don’t know how that information is interpreted or stored. We don’t know the equivalent of a computer language of the neurons (or other tissue types that may also learn). We know that we require input to learn, as do computers, the similarities kind of end there. That said, we definitely learn differently from how LLMs “learn.”

  • source
  • parent
  • hideshow 1 child comment
  • [–] -2 points 1 month ago (6 children)

    Counterpoint:

    “Teach” the LLM the proper procedure for counting the Rs and tell it to commit that to a file (or as a tool call). Then from there it will suceed 100% of the time.

    It doesn’t matter if it’s true learning or not. When you teach an employee to do something you give them the procedure and have them do it. If AI can be fed a procedure too, it will do it just fine even if it’s not in its true training. Does it matter if it’s consuming the procedure as prompt context everytime? Hell a lot of humans do read the procedure every time when it’s complex enough.

  • source
  • parent
  • hideshow 6 child comments
  • [–] 3 points 1 month ago (5 children)

    What the example shows is that you cannot "teach" the llm how to count the letter R, because LLMs don't work that way.

    The AI company would have to solve the problem another way, let's say by recognizing that the user is asking for letter-counting, and pass that off to a different kind of algorithm that can count letters.

  • source
  • parent
  • hideshow 5 child comments
  • [–] -4 points 1 month ago (4 children)

    Yes, and that’s called a harness, which everyone uses these days. The harness increases perceived intelligence (or accuracy) by absurd amounts. You can “teach” (or the equivalent of) LLMs anything with a custom harness.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 4 points 1 month ago (2 children)

    How is it teaching? Sounds like programming to me.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 0 points 1 month ago (1 child)

    Well the idea is you don’t need to code it yourself, you can have it do it for you. Sure it’ll have bugs the first few times, but humans also make mistakes the until they get the hang of it.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 4 points 1 month ago

    The difference i guess is that humans are capable of learning and producing better code as they become more expert at it.

    The big tech companies are surely trying to improve AI with these "harnesses" as you call them, and you can try vibe coding them yourself.

    But it seems to me like spending a lot of time adding features to a technology to try to catch edge cases, but the edge cases will never end, and you'll never be able to use it for anything except rough approximations or bullshit

  • source
  • parent
  • [–] 8 points 1 month ago (1 child)

    As you converse with it and it responds, part of the input is just your entire conversation up to that point, and its responses, plus your new prompt. That’s it that’s how the memory works.

    And for this you pay an increasing amount for tokens.

    Michael Reeves did an interesting piece on it: https://youtube.com/shorts/WP5_XJY_P0Q

  • source
  • parent
  • hideshow 1 child comment
  • [–] 1 point 1 month ago

    In my experience, it seems to be limited to previous conversations, and it may be limited in length.

    I pretty regularly delete old conversations, partly with the guide of privacy which I am sure does not actually delete or exist, but also for my own sanity of looking back on projects. And I don't even use AI that much, it's all free tier, mostly Claude.

    That said, it will often get "lobotomized" and "forget" things after I delete the source conversation.

    It does remember basic things. Like I use it a lot to help set up Docker containers, so it knows I have a Linux machine and a webserver and a Windows Machine. It knows the Linux box runs Docker and is named Squall. It knows these sort of, generic things, even across conversations. I can open a new chat and say "I want to set up a Wordpress sandbox" and it will ask like "Would you like to use Docker on Squall or run it outside a container with Apache."

  • source
  • parent
  • [–] 0 points 1 month ago

    Most harnesses that people use to interact with llms have memory and skills.

    You can "teach" them a skill which is sort of like a macro so you can say "deploy to dev" and it'll look up that skill and execute it without any extra input needed.

    For memory it'll keep files on your local of what it's seen before and can pull from the information in those files if necessary. So if you always ask it to label reports in a certain directory "Midwest expansion project" the harness will add that to its memory files and start to label it automatically.

  • source
  • parent
  • [–] 106 points 1 month ago (9 children)

    “Cooper AI Co-founder & CEO”

    Why would you expect her take to be anything but brain-dead?

  • source
  • hideshow 9 child comments
  • [–] 40 points 1 month ago (2 children)

    Exactly.

    Im pretty sure graduates aren't beating down her door for the opportunity to work for her.

    Sniffing her own farts too much.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 19 points 1 month ago (1 child)

    It's 2026 and yes, I can imagine that there are some rich college kids who think paying for an internship with her would be a smart move.

    They'd get a chance to build up contacts and see how she operates.

    Telling the Queen how good her farts smell has long been a path to success.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 15 points 1 month ago (5 children)

    I'm pretty underwhelmed before even clicking...

  • source
  • parent
  • hideshow 5 child comments
  • [–] 7 points 1 month ago (4 children)

    Because, as we all know, the only people who can be adequately replaced by the inference-free prediction machine are business idiots.

    Everyone else has a real job that actually requires real intelligence instead of prediction algorithms.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 3 points 1 month ago* (last edited 1 month ago) (3 children)

    Funny enough, though their jobs would be the easiest to replace, they are "important" in that they can be held legally responsible for business decisions (Responsible Corporate Officer doctrine). That's partly why they get tens of millions of dollars - so they can take the fall when shit hits the fan (either they fuck the shareholders or majorly break the law).

    This is known as the "Scapegoating Premium" - you are basically paid for the risk of taking on a very public, high liability/visibility position. The corporation pays their legal fees and has insurance to help with this, but it's compensation for the possibility of legal and reputational damage.

  • source
  • parent
  • hideshow 3 child comments
  • [+] 3 points 1 month ago* (2 children)
  • [–] 77 points 1 month ago (5 children)

    If I'm paying to start working for you, you'd better sign a contract guaranteeing my long term employment at a high salary. Otherwise, what's to stop you from laying me off the second I'm done paying you?

    I'm so tired of these grifters.

  • source
  • hideshow 5 child comments
  • [–] 39 points 1 month ago

    This will happen. In the UK the government tried to bring back slavery in the early 2010s. Unemployed people were told that they had to work for large corporations just to keep getting their unemployment benefits. It came to about £1.50/hour or so when minimum wage was significantly higher than that.

    Corporations got free labour out of it, tax payers paid the unemployment benefits for the benefit slaves. They often said you may get a full time job at the end of it. You won't, they will find someone else who is unpaid.

    Oh yeah, and some of those unemployed people used to work there but lost their jobs to unemployed people and then had to join them.

  • source
  • parent
  • [–] 4 points 1 month ago

    Done.

    PS, we are now also making sure you clock in and out on the dot and tracking your keyboard presses per minute for productivity and reserve the right to use any of these things to terminate you for being a "bad" employee.

  • source
  • parent
  • [–] 61 points 1 month ago* (last edited 1 month ago) (7 children)

    What a valuable contribution from the co-founder and CEO of "Cooper AI"... on a platform for business-type show-offs who have nothing better to do than reinforce each other’s outrageous incompetence...

    So this is the scope of "Cooper AI’s" operations:

    Executives Who Use Cooper Make Better-Informed Decisions. Cooper uses natural language to generate insights and deliver the high-impact opportunities your executive team needs right now.

    I think it’s pretty obvious from this sentence alone that this is nothing more than another front-end that forwards all requests to the actual cloud LLM providers.

    It’s truly absurd that there are people who see any added value in this.

  • source
  • hideshow 7 child comments
  • [–] 5 points 1 month ago (2 children)

    "Generate insights" had me rolling my eyes. If she thinks that's what they're doing, she deserves everything that's coming to her.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 52 points 1 month ago (7 children)

    Pay peanuts, get monkeys. AI hasn't changed that equation, so if they think that a grad will accept negative pay and come out the other end skilled, they are on crack.

    Businesses have always paid to upskill young employees, its a cost of being an employer.

  • source
  • hideshow 7 child comments
  • [–] 11 points 1 month ago (3 children)

    The problem though is that AI reduces the number of places to work at which will increase the competition between applicants. If some employers start doing this there will be applicants that will pay to get ahead. How big of an issue this becomes depends on how big of a competition AI will place on job seekers.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 13 points 1 month ago* (last edited 1 month ago) (2 children)

    I don't think AI changes that balance all that dramatically. Any business that won't hire juniors now, wouldn't have hired them before, and simply had the "Must have 20+ years experience in [flavour of the month] framework" in their job ads.

    And experience from past jobs hasn't really counted for anything beyond "They werent stupid enough to get fired", so paying for experience doesn't make sense either.

    All that said, regardless of who has the better crystal ball, its a shit time to be a compsci grad...

  • source
  • parent
  • hideshow 2 child comments
  • [–] 5 points 1 month ago (1 child)

    hasnt CS majors had low job prospects for a decade+ already, even back then there were talks on other forums how they cant find jobs.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 5 points 1 month ago* (last edited 1 month ago) (1 child)

    They used to, until many realized that there's no reason to train fresh grads if you can just poach then from another company after they've been trained for a while and if you train your own, they will get poached similarly by someone who wants to skip the training. They've been treating it as a zero sum game.

    Increasingly few companies will hire fresh grads now. Naturally, that causes a shortage in people with experience. Not to worry, they're now complaining about the lack of potential employees with experience and demanding the government to invest money in advertising our country to third world devs and to let them hire foreigners more easily instead of having to look for people here first. So now part of my tax money goes to an initiative to make sure companies can keep not training young devs.

    This had been brewing for years now and I'm disgusted.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 3 points 1 month ago*

    it was the law of diminishing returns, its the same for bio/biotech. they wanted experienced grads by graduation time, so if you are unable to get into a lab before. your chances are low. employers trying to save money by not "training a entry level/junior" and expecting them already have years of experience upfront. worst now its skewed heavily(rfor bio) gender wise.

  • source
  • parent
  • [–] 49 points 1 month ago* (last edited 1 month ago)

    There is NO legitimate reason to endure the fresh graduate.

    So many incredibly entitled people have never felt the obligation to give to others that which was given to them. These people should be shunned by civil society.

  • source
  • [–] 40 points 1 month ago (1 child)

    What an absolutely unhinged shit take. Even at the most generous interpretation, LLMs are at the capability level of a stupid intern you always have to check and correct, and it doesn’t learn (possible exceptions is narrowly defined problems spaces like coding)

  • source
  • hideshow 1 child comment
  • [–] 31 points 1 month ago (2 children)

    Slopvangelical LLM thumpers are proud of their acquired brain injuries, and believe everyone should be equally enhanced.

    The opinion of anyone impressed with the function or output of a LLM, to the level they believe it is in any way adjacent to intelligence, should be summarily dismissed, and if they persist, ridiculed.

  • source
  • hideshow 2 child comments
  • [–] 23 points 1 month ago

    My LinkedIn feed is overflowing with executives unabashedly posting their navel-gazing chats with bots.

    "What would you look like?" "What is it like to be you?"

    They're just so excited to finally have a friend at work, I think.

  • source
  • parent
  • [–] 8 points 1 month ago

    proud of their acquired brain injuries

    This here is hurting more than it may look at first sight when you have children who fall into that category. Ask teachers, they know that kind, and they fear for that kind.

  • source
  • parent
  • [–] 29 points 1 month ago

    Capitalist brainrot.

    That said, coop programs exist.

  • source
  • [+] 29 points 1 month ago (11 children)
  • [–] 2 points 1 month ago

    She's running an AI startup to service insurance companies, she's like 3-4 levels removed. Probably a strong case for conservative government grants except she's a womand and a minority

  • source
  • parent
  • [–] -1 points 1 month ago (8 children)

    The answer is kind of hinted.

    How do you pay for education if you don't have a job right now? Universities don't pay their students, don't they?

  • source
  • parent
  • hideshow 8 child comments
  • [–] 0 points 1 month ago (7 children)

    lol entry level jobs are not university level education. the fact that someone had to type this out to explain it to you says loads about you

  • source
  • parent
  • hideshow 7 child comments
  • [–] 1 point 1 month ago (6 children)

    Of course they are not, did I claim that? Or are you talking with some strawman arguments in your head?

  • source
  • parent
  • hideshow 6 child comments
  • [–] 1 point 1 month ago (5 children)

    How do you pay for education if you don’t have a job right now? Universities don’t pay their students, don’t they?

    just replying to the idiotic shit you wrote bro.

  • source
  • parent
  • hideshow 5 child comments
  • [–] 0 points 1 month ago (4 children)

    And you are really showing your inability to comprehend it

  • source
  • parent
  • hideshow 4 child comments
  • [–] 1 point 1 month ago (3 children)

    LOL You can't even explain what you meant. how silly and sad are you.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 0 points 1 month ago (2 children)

    Oh, I am sorry, I was just not aware that you asked for explanation, you just sounded like you wanted to be rude. If you actually capable of having decent conversation, I'd be happy to. Do you though? Or

  • source
  • parent
  • hideshow 2 child comments
  • [–] 1 point 1 month ago (1 child)

    you couldn't explain it with a week and cliffs notes, because it's a stupid idea.

    asking people to pay for internships is dumber than a box of rocks, and your continued enthusiasm for the scheme makes me wonder if you're a linkedin-lunatic, techbro, or ai schlorping asshat.

    at the end of the day I don't care, fuck off dweeb.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 27 points 1 month ago* (last edited 1 month ago) (3 children)

    And by the way, every talking head out there saying that the future rests on teaching our kids how to use AI properly: puh-lease, that will come naturally, it's like saying teach your kids how to use the internet.

    Allow me to introduce you to David's Law, the legislative result of not teaching children the dangers of the Internet.

  • source
  • hideshow 3 child comments
  • [–] 26 points 1 month ago
    [–] 26 points 1 month ago (1 child)

    People will never be free until the last CEO is strangled with the entrails of the last linkedin lunatic.

  • source
  • hideshow 1 child comment
  • [–] 25 points 1 month ago (1 child)

    Companies would totally force employees to pay for their work, if they could get away with it.

    1. Doesn't forget what you said once you said it

    I'm betting she asks the chatbot whether she remembers something or not, which always returns with a dumb "Yes! Of course you remember!"

    that will come naturally, it's like saying teach your kids how to use the internet

    Considering the amount of kids who think the internet is just youtube, tiktok, instagram, xitter, roblox and whatsapp, yes, we need to teach them how to fucking use the internet.

  • source
  • hideshow 1 child comment
  • [–] 5 points 1 month ago* (last edited 1 month ago)

    Considering the amount of ADULTS who think the internet is just youtube, tiktok, instagram, xitter, CS and whatsapp, yes, we need to teach them how to fucking use the internet. :/

  • source
  • parent
  • [–] 25 points 1 month ago* (last edited 1 month ago)

    As a side note, you should be teaching your kids how to use the internet

  • source
  • [–] 24 points 1 month ago (3 children)

    Or, we could just give everyone UBI and admit that capitalism is a failure?

    Stop worrying about profit and just let people have enough to be happy?

  • source
  • hideshow 3 child comments
  • [–] 12 points 1 month ago* (2 children)

    Just like there is no crying in the casino, there is no logic or empathy in the mental illness epidemic of fascism.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 6 points 1 month ago (1 child)

    A few years ago I was chatting with a co-worker. We had the exact same job. I asked him what he would buy with a billion dollars that he couldn't get with $50 million.

    He couldn't name any specific thing he'd want, but he was enraged at the idea that $50 million was the same as a billion.

    No idea what to do with the money, but he was sure it was 'better.'

  • source
  • parent
  • hideshow 1 child comment
  • [–] 24 points 1 month ago

    Controversial Sewage-brained take:

  • source
  • [–] 23 points 1 month ago

    She'll be looking for sympathy after the a.i. bubble bursts and she's bankrupt.

  • source
  • [–] 23 points 1 month ago*

    No evidence, just mindless opinions based on nothing.

    Here's my take: In the future, Tomatoes will become sentient from genetic engineering and will wish to overthrow their masters. Only gen z, alpha and AI will be able to stop this....Attack of the Killer Tomatoes. I'm smart and realistic.

  • source
  • [–] 22 points 1 month ago (2 children)

    Excuse me, haven’t we already seen tech companies going back to using junior programmers and boilerplate code because the LLM tokens are more expensive than just hiring people?

  • source
  • hideshow 2 child comments
  • [–] 9 points 1 month ago

    That's why taxpayers are paying for massive data centres that poison their neighbourhoods.

    More data centres means token become cheaper and more jobs get replaced.

    Tax payers literally paying to reduce the number of jobs for their children to work in the future.

  • source
  • parent
  • [–] 7 points 1 month ago

    haven’t we already seen tech companies going back to using juniors

    Not really no. There might've been few outliers.

    Companies are trying to spend their money more efficiently right now (as in, "make the genie work peasants, but for pennies").

    Source: ehh, I still work for one.

  • source
  • parent
  • [–] 19 points 1 month ago (2 children)

    HOTL > Human In The Loop

    Jenny, you have one job.

    I think the CEO mean Chief Executive Idiot.

    I felt bad for your ancestor, that my ancestor can feel the second hand shame.

    I'm out of insult for now.

  • source
  • hideshow 2 child comments
  • [–] 19 points 1 month ago (4 children)

    This is monumentally out of touch with how universities in the US work these days.

    One of the main reasons I choose the university I did was the co-op program - basically curriculum-mandated paid internships. Every other semester after freshman year, I was working a real engineering job at a real engineering company, and by the time I graduated, I had 1.5 - 2 years of actual industry experience and a budding professional network built out. And tons of universities offer programs like this, with more implementing it every year. Try telling a student from one of those universities, "no, see, actually, you have to pay me to work this entry-level job you're already overqualified for". Lol. Lmao, even.

  • source
  • hideshow 4 child comments
  • [–] 7 points 1 month ago

    This. I was required to find a job in the industry over the summer or I was not invited back to college the next year. If you couldn't find a job/ internship/ or summerstock the faculty would help you find something or, if you were the right person just give you a job because the entire staff was also required to work professionally.

  • source
  • parent
  • load more comments
    view more: next ›