Reply With QuoteAn AI with access to worldwide language which it drew from perhaps? The images certainly align with how I've seen AI interpret them.
I saw that "AI" generated pizza commercial and wasn't sure it's legit.
"Are you ready for best pizza of life?".
Straight up, that's NOT the type of thing ChatGPT would even output. It should be very rare to have the word sequence "for best".
The thing feels fake, because it includes grammatical errors to make it feel more "AI" but they're not even the type of errors that AI tend to make.
AI don't just randomly forget to have participles and pronouns. However that's common with some non-English speakers, whose native languages don't have those joining words.
So, a Russian speaker might conceivably say "Are you ready for best pizza of life?", but I can't see how the hell an AI would come up with that.
So it feels more like some human just removed words like "the" "of" "for" and "your" to make it feel more "robotic".
Last edited by Cipheron; 25th Apr 2023 at 22:33.
An AI with access to worldwide language which it drew from perhaps? The images certainly align with how I've seen AI interpret them.
No need to speculate, here's the dude's twitter post on it. Inasmuch as you can believe anything, anywhere.
Well the guy says GPT4, so you'd assume that's pretty advanced.
Someone else actually called him out on the dialogue being too choppy to be GPT and the creator claims he deliberately asked GPT to write it poorly. Which is definitely possible.
Yeah what's kind of funny about that is that, unlike a few years ago when those memes were going around (like "I trained an AI on 50 hours of Family Guy episodes and it wrote this scene") which pretty transparently were written by a person to sound like an AI, today ChatGPT is good enough that there's no reason to even bother with the subterfuge ... because you can just ask ChatGPT directly to sound like what people think cheap AI should sound like.
The catch I guess is that it's developing so quickly that a lot of people may still not even realize just how good it is, and they're still thinking about it like those old memes. So I could see it happening either way with this guy, either he actually did an oldschool style meme (by himself in its voice) without realizing ChatGPT is so good he didn't need to and he just came up with that story as a post-facto cover for his ignorance or whatever, or he knew how good it was and deliberately asked it to sound like the oldschool style meme. I actually don't care enough to get to the bottom of that mystery, only just enough to write this post. =V
My biggest worry about ChatGPT is that people are more and more starting to rely on it as a source of knowledge and problem-solving. Stack Overflow banned it several months ago already due to being overflooded with ChatGPT generated garbage answers and I've been seeing a similar reaction in language learning communities. It can generate useful and generally good enough explanations, but it's often subtly wrong in ways you can't really catch unless you already know the answer very well and the problems really start when you get to questions that aren't straightforward enough for them to be a good amount of text on it available on the internet or that haven't been prominently answered multiple times. This is an issue even very sophisticated algorithms like ChatGPT can't overcome -- it can only ever be as good as its source material, and usually it's less good.
Systems like this can be used to give someone with at least moderate skill in an area ideas for alternative approaches, but low (and no) skill users will struggle to understand why the answers aren't useful on their own.Originally Posted by Starker
Sure, this is the best case scenario -- a user that already knows what they are doing and has plenty of skill to accomplish a task by themselves uses it to generate a boilerplate/mediocre solution that they can then independently verify is good/accurate enough. I'm not really worried about that especially since it has already been happening pre-ChatGPT, where people would use scripts/code written by other people and substitute their own variables or plagiarise an essay off the internet and change the wording slightly.
The main issue with low/no skill people using solutions presented by ChatGPT (and similar) is trying to slot those solutions into something where they have potential to cause actual harm.
Moderately/highly skilled people using systems like this to improve their workflow is generally beneficial as long as they don't blindly trust what is offered.
There is also the problem that the sheer volume of algorithmically generated texts like these massively outweighs the number of people available to adequately assess these texts.
Also, from what I keep seeing, quite a few people seem to think that these texts are somehow created with some sort of an intelligence behind it (probably because it's being marketed as an AI), so they take them far more seriously and use them far more carelessly than they should.
I guess this must be what it was like when Eliza first appeared.
With Eliza, the cracks in the illusion usually showed within one or two exchanges. It was never particularly convincing, unless you fabricated an example playing to its limited strength. No comparison to what you can get out of ChatGPT, at least in terms of surface convincingness.
I think the point being made was about people seeing more than is actually there, and how this goes back to the very earliest and most primitive chatbot ever made.
It's not even sentience, people wrongly ascribe some "decision making" process to many chatbots, i.e. there's some algorithm in there that's using logic to decide what to write.
When in fact it's actually a deceptively simple "word salad machine" that's just been trained to fool another computer program as to whether the word salad is fake word salad or from real texts. The thing is, neither the program generating the word salad, nor the training program that assesses the output understand English or the topics it's being trained to produce texts about.
Exactly. I remember in particular the anecdote where the secretary of Eliza's author asked him to leave the room so that she could have a real conversation with Eliza. Maybe it's because the answers ChatGPT generates are remarkably human-like, but people can't seem to help but anthropomorphise it or at the very least think it has a brain of "some sort". Like, saying that it can "come up" with things or "create" something. And I'm not saying it to disparage anyone, as I myself have been compelled to do the very same, especially before I read up on how it actually works.
Just to put on my old philosophy of language cap for a bit, while it's not directly simulating the brain, it's effectively recreating the same representations the brain would have to use to get the same outputs in the same context. So I think it's still fair to say it's internalizing human ways of thinking, not in the way humans do it, but in its own way.
But it's not able to recreate everything. But I think that's mostly just because it didn't have access to other modalities of experience, like vision, audition, or all the emotional and affective inputs that human brains trigger instinctively, not in the training and not in the chats. But if it did have those inputs, and probably you'd need a lot more density in the hidden units to handle that, then I think it'd internalize how the brain represents those things as well, at least to the extent that inputs match the outputs in the way training structured it, no more and no less.
I think the argument I'm handwaving at is something like, while I agree it'd be wrong to over-anthropomorphize ChatGPT, I think there's a problem with under-anthropomorphizing what it's actually internalizing from human behavior in its hidden units also. It's very different from Eliza & GOFAI (good old fashioned AI), which didn't represent the functionality of "meanings" at all, just brute-force stim/responsed in "blind" data arrays. Hidden units aren't just blind data arrays; functional representations are being built in the hidden units. They're opaque and we can't see them in there, but that doesn't mean they aren't there.
That's the thumbnail version. I think I'd have to spend a lot of time unpacking what I think I mean by that to give it justice though.
I don't really know how we use language on the brain level, but I know we use it extremely creatively. We are constantly playing around with it in the Wittgensteinian sense of language-game. As such, any meaning is infinitely malleable and depends on the current game being played. And not only that, we even change the rules mid-game and invent whole new ones as we are playing it.
We certainly don't go, "Based on the accumulated texts I have read, this is statistically the most plausible word I should use in the sentence I've been saying so far. Now, let's run a lot more probabilities and check which one of these different end results best corresponds to the user prompt according to my training data and gets the best reward from the reward model."
Because it's the last bit that gives GhatGPT the most human-like appearance, that the output has been carefully selected to appease human trainers and weighted towards the most plausible-sounding ones.
Re: [[We certainly don't go, "Based on the accumulated texts I have read, this is statistically the most plausible word I should use in the sentence I've been saying so far.]]
We certainly do exactly that on the first pass (well not exactly but virtually), which actually gets put out when we're not paying attention or are distracted, etc. (A: "Hey, what's going on?" B: "Fine, and you?"). When we are paying attention, there's a check on the statistically-likely-reply that's first put forward against the context & our goals, etc. That also happens with what we hear other people saying to us, cf. the famous N400 event in EEGs (a negative polarization bump 400 ms after you hear a thing) when a person's comment diverges from what we expected them to say in the statistically-likely-model that first gets pushed out within the first 100~200ms.
But I wasn't even talking about that. What I was talking about is that functional features may be getting built into the hidden layers via the way back-training works on the feed-forward models. It's like when you train a model to identify faces, operations necessary to discriminate faces like discriminating sub-features like lines and curves actually get built into the model. Or representations of the sub-features are getting burned into the model, so the vanilla ML algorithm "operates" on them. It's like the way natural selection selects fit genes inside DNA code that "operated on" give you fit phenotypes, back-training selects weights in feed-forward ML hidden layer code that "operated on" gives you cognition. Yes it's still a statistical model, but it's effectively simulating operations of line and curve discrimination inside the guts of that model.
Or back up; you're saying "it's just computing statistics blindly". Literally it's a gradient descent algorithm to a statistical nadir (a peak flipped on its head). But if the path of the descent itself effectively simulates the mechanics of cognition, then that "blind path" also happens (not coincidentally) to "operate the mechanics of cognition" in a gradient descent packaging that statistically lands on the same statistically likely peak/response because, after a billion training runs blindly carve at the geometry, that geometry is going to always converge on to the only "statistical" path that's going to consistently get you to the right peak, and that's going to be by actually baking a simulation of the cognition required to get you there into the path geometry itself.
You're actually building algorithms of cognition into the geometry of the model. That's the argument. So "just computing the statistics" is also simulating something like what the brain does, at least the parts that link inputs to outputs. But you're not seeing it because you're not looking at it as actual operations being done by the gradient descent path as it winds its way around. You're just seeing a gradient descent to a nadir and calling that "statistics", a blind road that happens to statistically land on the right answer. (I mentioned lots of little things that aren't getting baked in because there aren't enough cues in the outputs to bake them in. But if they're not making a difference to the output at all, you'd have to look at what role they're playing at all. That's a worthwhile thing to follow-up on, but that's another discussion.)
Since it is so opaque what's happening in the gradient descent space (it's this vast dimensional space we usually just bracket and dismiss as random meandering), there's room for people to argue what's actually happening down there. I think there are a lot of smart people that would disagree with this take, so I don't want to say it as if it's some consensus position. For the purposes of this post I'll just say this is one way that some people look at it that, to the extent it's on to something, gives a view where ChatGPT has more human-like thought inside it than people are giving it credit for.
And the thing about that argument is, the argument isn't really that baking statistics into the geometry of a gradient descent that ChatGPT does can simulate the cognition like a brain does. The real argument is that the algorithms of cognition like the brain does are made by baking them into the geometry of gradient-descent-like operations that the brain's neural networks are doing that are a lot more like what ChatGPT is doing than people think. There are still big differences, I still agree with that; and to get at some of them I'd have to talk about, e.g., Grossbergian neuroscience (e.g., the models are dynamic in the brain whereas they're static in ML/ChatGPT, etc.), but the differences aren't the kind I think people are thinking about when they say ChatGPT is nothing like the brain.
All of this is again just an idea I'm throwing out there as food for thought.
---
Edit: I'm not fully responding to your response though. ChatGPT doesn't answer like it's playing a Wittgenstenian language game because it's often not playing a language game with its responses the way it's set up, at least not as much as humans do. (Some of that also gets baked into its model whatever query it's responding to, I think.) But I think that's more to do with how it "understands" its role in giving responses to queries than the structure of the model itself. It probably would be even more human-like if it was more set up to take on the role of "acting" or "playing a game" in its responses than it does now. I mean, I think that's not as fundamental a change as one might think, like the difference between a bot that explains what a game bot would do in a map vs. an actual game bot playing in a map.
Last edited by demagogue; 28th Apr 2023 at 18:42.
I think it's more deeply different than that.
1. When humans write there's more going on, in between each word. With ChatGPT there just *isn't*.
This explains examples where ChatGPT was asked to add up a bunch of numbers such as "13+63+22+144+55+37" and gives the wrong answer. Each time you regenerate the result, it would give a close, but effectively random answer.
It's asked to give the result of that operation, so it just pushes the whole thing into it's statistical-analyzer, and that gives it a probability cloud of possible next tokens. It then picks one of the tokens randomly. This in no way resembles how a reasonable human would approach the problem.
However if you give it the same task, but tell it to add them up one at a time, then it would output the partial sums, and give the RIGHT answer.
Now, OpenAI patched this *particular* exploit, probably by hard-wiring in something that would detect sums and manually insert the correct answer, but it effectively shows that, inbetween words, ChatGPT isn't actually capable of doing logic that takes more than a single step.
2. Humans are able to "look ahead" more than one word when thinking about which word to use. ChatGPT basically is incapable of doing this, at all. So it lacks ANY sort of overall directed or goal-driven stuff to dictate what it decides to write. The very next word is purely dictated by the previous words plus some dice-rolls.
3. The human brain has "state" other than the words it just wrote. ChatGPT doesn't have this, it only has the words it just wrote: that's it's entire memory.
These three things are all connected. Lack of any state or higher functions explains the attempt to just assign results of math calculations at random, and the lack of "look ahead" ability when planning out a sentence.
So, saying the human brain is also a neural network like ChatGPT. I think that's not really a helpful way to understand how this works and the relationship to a real person trying to write something.
Last edited by Cipheron; 29th Apr 2023 at 05:04.
Just got back onto the ChatGPT subreddit. Immediately hit by the sheer amount of people not understanding how AI works.
The latest person is upset because he asked Bing AI to summarize some videos and it didn't get it right. Like, they expect GPT to watch videos now, and understand what's going on?
Seriously, these people really do think it's like the Oracle of Delphi.
A few points I want to make.
1. About what I was arguing above, I'm not contesting any of the criticism Ciph gave to it. There is obviously a ton of capacities still missing from ChatGPT. I was trying to make a weaker philosophical claim, something like that "mental representations" tend to converge, which is just a claim or hypothesis I'm making based on stuff I've read. It's like how eyesight keeps independently evolving in a lot of different species under natural selection pressure; I think a mental representation in an LLM can start to converge to the complementary representation in the brain with feedback weighting. That's still very far from saying LLM's cognition is "like" humans; maybe at most that it's more a difference of degree (orders & orders of magnitude) and application (not combining vision & language, no memory, etc.) than type per se. So I don't think I'd deny anything about the response to what I posted. But to push back a little maybe, there are humans that are also missing vision and memory, and we can say they aren't neurotypical, but we wouldn't say what they think "the cat ran across the yard" means is too fundamentally different. It's just that the way LLMs like ChatGPT are designed right now, they don't have the resources available to actually act on that meaning, like bona fide free action, where they can actually imagine the cat and the yard, have some emotion or motivation based on it, and take some actions based on that motivation. I think the core of the representations are in there, but they haven't figured out how to use them like humans do. That's just my feeling about where things are now or will be soonish.
2. Regarding what was just posted above, I think the next step is going to be chaining AI. We saw it with that monk video which chained I think three different systems together--converting player voice to text, feeding the text to ChatGPT with appropriate pre-instructions, and the script or whatever in the game that parses the ChatGPT text and converts it to the speech and animations of the NPC. I think the people designing Bing AI and the like are going to starting chaining AI like that to start addressing questions like the above, because it's natural people will think of AI as doing those kinds of tasks, and they'll design the systems to the demand.
3. The trend I've really been noticing is how many videos are AI scripted and voiced, and how apparently aggressively these kinds of videos are being pushed on to me in different platforms. The way the commenters take it so for granted makes me think that the trend here is that our world is about to become universally and ubiquitously curated and commentated on by AI, where things happening just get automatically processed in this endless stream of AI updates constantly linking us to "the world" and giving us calculated opinions and questions that get people automatically engaged. It's something along those lines. I don't like the trend. I feel like there should be some resistance to this constant AI curation popping up, like some push for "authenticity", but I'm not sure there is or will be. I guess we're starting to see that in the visual arts and maybe music where people are taking sides, but I feel like even those are exceptions that prove the rule and are going to be submerged by the flood of AI content coming.
Barely 6 months after it gets released, there's already what looks like hundreds of books on doing everything under the sun with ChatGPT
Carl Sagan transcript:
Rest here: https://pastebin.com/LTMs1GFHIn the vast cosmic tapestry of existence, amidst the grandeur of galaxies and the ethereal dance of celestial bodies, there exists a humble but integral part of our human experience—a part that connects us to the very fabric of the universe. Today, I want to share with you a deeply personal reflection on a most mundane yet profound event: my own personal experience of releasing waste, of defecation.
As I sat upon the porcelain throne, contemplating the mysteries of the cosmos, I marveled at the intricate processes of digestion, the enzymatic ballet orchestrated within the inner sanctums of my body. And there, amidst the ebb and flow of peristaltic waves, the journey of sustenance culminated in the expulsion of a remarkably unique entity.
...
the first iteraction was as a Tom Scott video, it feels pretty on the money:
https://pastebin.com/DcA3jgPK
but you can't be sure if that's actually ChatGPT getting it right vs my confirmation bias.
so i generated a bunch with different youtubers and personalities. All of them felt pretty generic or like ChatGPT could have just been guessing what was appropriate based on the topic of the channels. However, as soon as I asked for Carl Sagan, it was clearly a VERY Carl Sagan response.
Last edited by Cipheron; 10th May 2023 at 05:26.
That's honestly the perfect average use case for ChatGPT at this moment, would that it would be limited to merely parodying science personalities well enough to get a chuckle out of me. I guess we'll just have to live with the infinite mountains of weaponised informational trash we designed to internet to deliver to our eyeballs so efficiently instead; though I would think it's about time we started working on popularising tools to deal with that, too.
Checking ChatGPT's skill versus a logic puzzle that requires lateral thinking:
... I was impressed when it churned this out really quickly, and it looked convincing at first glance, however, I picked up some errors in its logic.Prompt: Act as an expert puzzle solver Let’s say that you have 25 horses, and you want to pick the fastest 3 horses out of those 25. In each race, only 5 horses can run at the same time because there are only 5 tracks. What is the minimum number of races required to find the 3 fastest horses without using a stopwatch? Explain how you would solve it.
To find the three fastest horses among 25 horses using only five tracks and without using a stopwatch, we can follow these steps:
Step 1: Divide the 25 horses into five groups of five horses each. Race 1: Run a race with these five groups, one group per track. Label the groups A, B, C, D, and E. Note down the finishing order for each group.
Step 2: Determine the fastest horse in each group. From the finishing order in Race 1, select the horse that finished first in each group (the winners of each race). We'll label these horses A1, B1, C1, D1, and E1.
Step 3: Run a race with the winners from each group. Race 2: Place the five winning horses from Race 1 (A1, B1, C1, D1, and E1) in a new race. Note down the finishing order.
Step 4: Identify the top two horses in Race 2. From Race 2, the horses that finish first and second will be the two fastest horses overall. We'll label them X1 and X2.
Step 5: Determine the third fastest horse using the remaining horses. We know the horses that finished first and second in Race 2 (X1 and X2). We also have the horses that finished first in each group (A1, B1, C1, D1, and E1). So we need to find the horse that finished second in the group containing either X1 or X2.
Step 6: Run a race with the remaining horses. Race 3: Select the group that contains either X1 or X2 (let's assume it's group A, but it could be any group containing one of the top two horses). Run a race with the remaining four horses in that group (excluding A1). Note down the finishing order.
Step 7: Determine the third fastest horse. From Race 3, the horse that finishes first will be the third fastest horse overall. We'll label this horse Y1.
Finally, we have the three fastest horses: X1 (first place in Race 2), X2 (second place in Race 2), and Y1 (first place in Race 3). This process requires a minimum of three races to determine the three fastest horses among the 25.
It says to take the first two horses from Race 2 and those are the two fastest overall. But what it's failed to take into account is that the horse which came 2nd in Race 2 hasn't raced against any of the other horses in the group that had the fastest one - all of those horses might be faster than it. and what it's calling the 2nd fastest might be as low as 6th fastest overall.
This shows that ChatGPT is just spitting out plausible text and hasn't really understood the problem to be solved, or applied logic to solve it.
I could get it to output the accurate results, but only by baby-sitting it and specifying the right answer, and having it mimic me, rather than by explaining the logic and having it understand enough to work it out.
1 1 was a race horse.
2 2 was 1 2.
1 1 1 1 horse race,
and 2 2 1 1 2.
BTW i shared the result on Reddit, and someone claimed I just didn't prompt it correctly, and provided their own version.
Only it completely failed at picking the correct horses needed in the last race, and they didn't bother to check. Which was ironic because i mentioned in the post how it showed the importance of checking the results ChatGPT gives you.
Posting Permissions