You can't predict the next token in an arbitrary text unless you are highly intelligent and have a vast body of knowledge.
They're obviously intelligent in the way that we judge intelligence in humans: we pay attention to what they say. You ask them a question about an arbitrary subject, and they respond in the same way that an intelligent person would. If you don't consider that intelligence, then you have a fundamentally magical, unscientific view of what intelligence is.
To return to an analogy I used a couple of days ago ... birds can fly, planes can fly, ergo they are both flying things ... but they fly in completely different ways. So on the one hand (visible behavior) they are similar (or even the same), and on the other (physical mechanism) they are not similar at all.
Which one of these comparisons you want to use depends on context.
The same seems entirely possible for current LLMs. On the one hand they do something that visibly seems to to be the same as something humans do, but on the other it is possible that the way they do it entirely different. Just as with the bird/plane comparison, this has some implications when you start to dig deeper into capabilities (e.g. planes cannot fly anywhere near as slowly as birds, and birds cannot fly as fast as planes; birds have dramatically more maneuverability than planes, etc. etc).
So are LLMs intelligent in the same way humans are? Depends on your purpose in asking that question. Planes fly, but they are not birds.
To extend your analogy, imagine that there are airplane skeptics who insist that planes can't fly, will never fly, and are good for nothing. They only crudely simulate flight. Meanwhile, millions of people are flying around every day in planes.
But if by "flight" you meant "the sorts of things swallows and kestrels can do",then the movement of planes through the sky would be at best irrelevant.
This is simply wrong, and missing the point, simultaneously.
Flight (like "intelligence") means more than one thing. Planes fly, birds fly, but they not only use a different mechanism, they can't even do the same kind of flying that the other does.
Sometimes, the difference doesn't matter. Sometimes it does. Same for "intelligence".
We don't actually know that much about how the brain works, and nobody discussing intelligence will decide tomorrow that humans aren't intelligent if the details of how the brain functions turn out to be slightly different from what we previously thought.
LLMs obviously display what everyone prior to 2022 would have called "intelligence," before the goalposts started rapidly shifting with the release of ChatGPT. They can carry conversations about arbitrary subjects, understanding what you're asking and formulating thoughtful answers at the level of a very smart and extremely well educated human. They're not identical to humans (e.g., they don't have fixed personalities), but they display what everyone commonly believes to be intelligence.
I know you're arguing with someone else, but I think it is getting sidetracked.
Whether or not LLMs are intelligent (I think they are more intelligent than a cat, for instance, but less intelligent than a human) isn't my argument.
My argument is that complexity in and of itself doesn't yield intelligence. There's no proof of that. There are many things that are very very complex, but we would not put it on an intelligence scale.
The EU consists of 27 different countries, with substantial practical barriers between their internal markets (even if it's one single market in theory). Often, only EU intervention can overcome those barriers. Otherwise, you end up with national fragmentation.
What people in Europe do very frequently is buying from online shops of another country, either because they do not find the product in a local shop or because of better prices.
What is needed is a card for online shopping that is valid in all Europe.
But it's not the card. Wero isn't doing anything new. It's just yet one more payment method to implement. Adyen, Stripe, Shopify and many other already support different local payment methods.
If you view all of math as just a set of logic games with the axioms as the basic rules, then there's nothing unnatural about complex numbers. Various mathematical constructs describe various phenomena in the real world well. It just so happens that many physical systems behave in a way that can be very naturally described using complex numbers.
The guy invented the path integral in his PhD thesis. He invented Feynman diagrams and figured out how to do finite calculations in quantum electrodynamics. Unless you're a perfect human being, please, cut him just a tiny bit of slack.
I understand not watching a 3 hour video before leaving a comment, but this is a disrespectful reaction to a very well thought out video by a professional physicist giving a nuanced opinion about Feynman's legacy. She acknowledges many times in the video that Feynman was a great physicist who deserved his Nobel prize. The central topic of the video is dissecting his public image and the many books published under his name that he did not in fact write, including Surely You're Joking and indeed the Feynman Lectures, as well as criticizing misogynistic behaviors celebrated in those books that has left a negative impact on the culture of physics.
(And also, "cutting him a tiny bit of slack" is pretty lax language considering the behavior being criticized includes beating his wife.)
This was really frustrating me. YT started recommending this channel and I could recognize the voice as an AI impersonation but had no way to know if it was at least reading something really written by Feynman. Eventually I concluded it wasn't, but there wasn't clear criteria under which I could report the channel. I'm not sure it's even against YT's TOS.
misogynistic behaviors were cultural at the time, I agree they're abhorrent but people are embedded in their culture. The same is said of Hitchcock, (as an example) and his behaviour was unacceptable by todays standards. We've come some way from that but still a way to go.
From the about the authors in the OP's link "Feynman was a remarkably effective educator. Of all his numerous awards, he was especially proud of the Oersted Medal for Teaching, which he won in 1972.". He probably didn't do a lot of the stuff he popularised, but that was what he did, it is a skill taking abstract stuff and making it coherent. I know when I did physics (in the 90's) many swore by his books, particularly for quantum, it was a bit of a secret we'd have these incomprehensible books on quantum, and someone would say - have you seen "The Feynman lectures", they are good, I wish we had the videos available at the time.
> misogynistic behaviors were cultural at the time, I agree they're abhorrent but people are embedded in their culture.
Moral relativism is a thing, but I think a more useful way to think of it rather than just saying "misogyny was a thing back then, should we care he was a misogynist then?" is to ask "if he were to have lived and worked in the 2000s, would he associate with Epstein?" And to be honest… Feynman does strike me as the kind of person to have the intellect to attract Epstein's attention and also the, for lack of a better term, party attitude to go to a couple of Epstein's parties that would result in him having awkward press releases trying to explain that he just had no possible idea that Epstein was doing anything sexual with children and conveniently forgetting all the times he was on the private island for some party or another...
That's the real strong vibe I get from Surely You're Joking. He's the kind of person who wants to be seen as someone who gets up to wacky hijinks, to be seen as "cool," and he specifically interprets "cool" in a way that's misogynistic even at a time (when he was dictating the stories that led to Surely You're Joking) when misogyny was starting to become a professional hindrance.
(And one of the things that really worries me about Surely You're Joking is that it's often recommended as a sort of "look at the wacky hijinks you can get up to as a physicist," so recommending the book is a valorization of his wacky hijinks and... well, that's ultimately what Angela's video is about, that's a thing we need to stop doing.)
> That's the real strong vibe I get from Surely You're Joking. He's the kind of person who wants to be seen as someone who gets up to wacky hijinks, to be seen as "cool," and he specifically interprets "cool" in a way that's misogynistic even at a time (when he was dictating the stories that led to Surely You're Joking) when misogyny was starting to become a professional hindrance.
In my experience, everyone who says this is talking about exactly one chapter in Surely You're Joking, but they don't appear to actually have paid close attention to the story. It's a story that Feynman recounts about trying to pick up girls when he was younger. He was advised by an older, "cooler" man to be mean. Feynman tries it and it works, but he feels bad about it and says that he never did it again. People calling Feynman a misogynist for this story seem to have just skipped the end of the chapter.
It's been decades since I read Surely You're Joking, and I've completely forgotten about that chapter. It plays no part in my conscious recollection of the book.
The episode that really stuck in my mind was the episode about his competition with the abacus-user, who was better at math, which essentially ends with him giving up trying to explain how he could mental math a cube root faster, because the abacus-user was just someone who couldn't understand a math explanation.
I remembered enjoying the book, so having not read it in a long time, I tried sharing Surely You're Joking with my kids at bedtime.
That chapter wasn’t the only thing I ended up skipping or heavily editing.
* Picking a room at Los Alamos with a window facing the women’s housing, but being disappointed that a tree or something blocked his view. (Wasn’t he also married at this point?)
* Starting a new Uni faculty position and hanging out at student dances, dismayed that girls would stop chatting & dancing with him when they learned he was a prof and not a fellow student.
* Hanging out at strip clubs to practice his drawing skills.
* Considering a textbook sales rep’s offer to help him find “trouble” in Vegas.
So maybe that one chapter turns around some at the end, but it’s not the only cringe-worthy moment in the book, and I can see why some people may have an overall negative opinion.
If I were going to do this with my kids now that they are teens, I wouldn’t filter as much and use the more questionable events as points of discussion.
This is from Lawrence Krauss[0]'s email to Epstein[1]:
> ps. I have decided that Feynman would have done what I did... and I am therefore content.. no matter what... :)
> On Apr 6, 2011, at 3:56 PM, Jeffrey Epstein wrote:
> what evidence? no real sex.. where is she getting her so called facts
Krauss's letter is obviously horrible in its implications. What's interesting to me is his interpretation of what Feynman would have done. Is it his delusional justification of what he'd done with Epstein, or is it based on a certain reputation of Feynman in the science community?
> misogynistic behaviors were cultural at the time, I agree they're abhorrent but people are embedded in their culture. The same is said of Hitchcock, (as an example) and his behaviour was unacceptable by todays standards. We've come some way from that but still a way to go.
The video actually addresses this very point in the first few minutes:
> the second component of the Feynman lifestyle that the Feynman bro has to follow, you know as told in this book, is that women are inherently inferior to you and if you want to be the smartest big boy physicist in the room you need to make sure they know that I think people are sometimes shocked to hear this like that that exists in this book especially because as I said if you were a precocious teenager interested in physics people shoved this book at you they just put it into your hands like oh you want to be a physicist here's the coolest physicist ever
> I feel like it's at this point in the video when like Mr. Cultural Relativism is going to show up in the comments and be like how dare you judge people from the past on their actions that's not fair things were different back then women liked when men lied to them and pretended to be an undergrad so that-- it was fine back then it was fine and I just, no, actually this book was published 40 years ago which is just not that long ago Richard Feynman should have known that women were people 40 years ago like absolutely not it's not "how things were back then" what's wrong with you people, no, it's inappropriate then it's inappropriate now
Later the actual author, Ralph Leighton, of "Surely You're Joking, Mr. Feynman!" is mentioned so perhaps the responsibility for what was included is his more than Feynman's. I think the criticism stands that the degree of sexism effectively celebrated by inclusion was certainly less culturally accepted in 1985 when the book was published than when the events occurred, and that's the point of raising the issue of why was it judged as good and proper to include this marginalizing anecdotes when his actual contributions to physics and teaching were worthy of celebration.
I do not think Feynman was celebrating his activity in the book. From memory, he learnt the behaviour from other bar flies at the bars he hung out. And he expressed his surprise at how some women reacted. This was far from his upbringing and his experience with his fiancee.
The behaviour is hardly laudable, but "celebrated" it is not.
> I do not think Feynman was celebrating his activity in the book.
The argument presented in the video about this is that these are the stories Feynman edited and reworked over time, and shared with his friend Ralph Leighton, who then recorded them in the "Surely You're Joking" book.
The video also describes a change in his behavior later in life. In 1974, responding to a letter asking to reprint "What is Science?"[1] from 1966, he comments that "some of the remarks about the female mind might not be taken in the light spirit they were meant"[2]. This is cited in the video as Feynman becoming more progressive between 1966 and 1974. The "Surely" book is published in 1985, and yet still includes the misogynistic stories. The video's complaint is that there should be some contextualization about views changing, like was given in Feynman's reply in 1974, but there being none it comes across as an implicit endorsement. I don't recall from the video if Feynman reviewed or edited the "Surely" book, which leaves it as Ralph Leighton's perspective more than Feynman's.
It seems a legitimate criticism that this book held up as an example of a good role model in physics doesn't try to avoid perpetuating bad stereotypes. It's probably egregious to say the mere inclusion of the stories celebrates their actions. But it's equally egregious to fail to even try to address the bad behavior, especially when it's held out as a positive example.
Certainly. But you're missing the point. Feynman chose to tell the stories to Ralph Leighton who then recorded them in the "Surely" book which was published in 1985, well after Feynman's own perspective seems to have changed about the more offensive things he'd said.
By many other accounts he was a kind, caring, thoughtful person, but some of the selected stories in "Surely" paint a significantly different picture. To me it's unclear, not having studied the life of Richard Feynman, what parts are exaggerated. But it does seem clear that these stories were refined and selected for inclusion, and were therefore considered endearing or representative for the intention of the book. And in the time and culture in which it was published that seems like a bit of a miss at the very least.
I've watched large sections of this video before, because it gets posted often. It's a 2-year-old video.
Based on that viewing, I think the author has a chip on her shoulder about Feynman, and is dismissive about his teaching and books, and is set on convicting him of being a very naughty boy.
One of the things that stand out from the video: The speaker says that Feynman didn't write the Feynman lectures. Wrong. He wrote and delivered the lectures. If you go to Caltech's Feynman lectures website, they even have audio of him delivering the lectures [0] and photographs of the chalk board [1]. How could someone make a 3-hour-long video about Feynman and not even know this?
Feynman was an immensely gifted physicist and one of the most (maybe the most) engaging and innovative physics teachers of the last century. You can criticize him for embellishing stories about himself, but those stories are incredibly entertaining and quirky, which is why so many people like them. He was a big personality, and it comes out in his stories. He wasn't a perfect person, but no one is, and there has been a movement in the last few years to try to demonize him (mostly unsuccessfully, given Feynman's continued popularity).
Finally, if one makes a video with a title like, "the sham legacy of Richard Feynman," one can't complain about getting pushback.
No, she's absolutely wrong about this. The book is based very closely on Feynman's lectures. He wrote the material and gave the lectures. Other people edited that material into book form, but Feynman did the lion's share of the work.
Saying that Feynman didn't write the book is just dishonest, unless you immediately clarify afterwards that Feynman did indeed write almost all of the material in the book, in something very close to its final form.
You should watch the whole video. From memory, the video author claims that the books are not based directly on the recordings nor on material that Feynman wrote himself, but rather on lecture notes written by another professor who had to cover for Feynman (who is also listed as one of the authors in the book). She also mentions how those lecture notes from this other professor correct some small mistakes Feynman made in some calculations and diagrams from the lecture. Her claim is that Feynman was not the person who actually wrote the text of the book.
You can literally listen to audio recordings of Feynman delivering the lectures. The book follows those lectures closely.
All lectures that professors deliver have mistakes in them. He produced a massive lecture series covering huge areas of physics over hundreds of hours of class time. There are bound to be typos and small math mistakes.
The complaint that Feynman had an editor makes me think the person who created this YouTube video has no idea how publishing works, not to mention academic publishing.
To me, claiming Feynman didn't write the lecture book is a stretch since they are fundamentally based on his lectures. But I think you are misconstruing some of her arguments and claims. I suggest you watch the whole video, because imo it does a good job at analyzing Feynman's figure.
I mean, for the most part the book is an edited transcription of what he said at the lectures (or, in some cases, what a guest lecturer said). But the lectures weren't scripted, and we know this because his lecture notes are preserved[0] and they do not contain anything like he full text of even a single lecture. They're just lecture notes, not a script. And of course, the book also contain a lot of example problems and graphics - those are mostly the work of Bob Leighton, I believe. There's a reason the book has had so many errata corrected over the years: it was never written and edited in the way a book manuscript would've been written and edited.
Now your complaint is that Feynman didn't literally write down every single word he was going to say? He prepared more than 600 pages of notes and then delivered hundreds of hours of lectures. They were transcribed and published as a book, with normal editing. Feynman is the primary author of that book, for good reason.
The one-shot performance of their recall attempts is much less impressive. The two best-performing models were only able to reproduce about 70% of a 1000-token string. That's still pretty good, but it's not as if they spit out the book verbatim.
In other words, if you give an LLM a short segment of a very well known book, it can guess a short continuation (several sentences) reasonably accurately, but it will usually contain errors.
Right, and this should be contextualized with respect to code generation. It is not crazy to presume that LLMs have effectively nearly perfectly memorized certain training sources, but the ability to generate / extract outputs that are nearly identical to those training sources will of course necessarily be highly contingent on the prompting patterns and complexity.
So, dismissals of "it was just translating C compilers in the training set to Rust" need to be carefully quantified, but, also, need to be evaluated in the context of the prompts. As others in this post have noted, there are basically no details about the prompts.
No one can really figure out what legitimate uses crypto has that can't be covered by normal payment systems.
Everyone can immediately see how useful AI is, and tons of people are using it. Pretending it will pass would be like saying the Internet was a fad in 1997.
Your productivity has not increased by a few times unless you're measuring purely by lines of code written, which has been firmly established over the decades as a largely meaningless metric.
I needed to track the growth of "tx_ucast_packets" in each queue on a network interface earlier.
I asked my friendly LLM to run every second and dump the delta for each queue into a csv, 10 seconds to write what I wanted, 5 seconds later to run it, then another 10 seconds to reformat it after looking at the output.
It had hardcoded the interface, which is what I told it to do, but I'm happy with it and want to change the interface, so again 5 seconds of typing and it's using argparse to take in a bunch of variables.
That task would have taken me far longer than 30 seconds to do 5 years ago.
Now if only AI can reproduce the intermittent problem with packet ordering I've been chasing down today.
I'm measuring by the amount of time it takes me to write a piece of code that does something I want, like make a plot or calculate some quantity of interest.
Or even the fact that I was able to start coding in an entirely new ML framework right away without reading any documentation beforehand.
I'm puzzled by the denialism about AI-driven productivity gains in coding. They're blindingly obvious to anyone using AI to code nowadays.
A few weeks ago I was interested in median price paid in the UK for property. I pulled down a 900,000 line csv from gov.uk and asked chapgpt to give me a python to parse it based on price (col 2) and county (col14), then output the 10,25,50,75,90 percentiles.
It dropped out a short file which used
from statistics import quantiles
Now maybe that python module isn't reliable, but as it's an idle curiosity I'm happy enough to trust it.
Now maybe I could import a million line spreadsheet and get that data out, but I'd normally tackle this with writing some python, which is what I asked it to do. It was far faster than me, even if I knew the statistics/quantiles module inside out.
I'm not adding a+b. It will be more like, "Calculate the following nontrivial physical quantity from this catalog of measurements, reproject the measurements to this coordinate system, calculate the average and standard deviation using this pixelization scheme, estimate the power spectrum, and then make the following 3 plots."
This would have taken me an hour previously. It now takes a few minutes at most.
I feel like many AI skeptics are disconnected from reality at this point.
It feels like our (US) political system; people in their camps refuse to believe any data proposing a benefit of the "other" camp.
For me, the rise of the TUI agents, emerging ways of working (mostly SDD, and how to manage context well), and the most recent releases of models have pushed me past a threshold, where I now see value in it.
reply