🗑️ Anonymous at Wed, 29 May 2024 19:55:25 UTC No. 16199510
>>16199497
This is sort of true but also not exactly true.
Lookup tables are discrete set mappings, whereas in general, a neural network is a smooth function. In fact neural networks are so smooth that they are literally universal function approximators for any Lipschitz continuous function in some R^D.
Depending on the architecture for the NN and the particular activation functions used, you might have significant "saturation effects" which would make your NN behave in a more "tabular" way rather than as a smooth function, but in general NN's are smooth not discrete. In fact that's a major reason why they are so scalable compared to tabular search methods, they don't need to search through discrete options and can instead essentially just do a bunch of very computationally arithmetic instead and arrive at an inference.
Anonymous at Wed, 29 May 2024 20:01:32 UTC No. 16199526
>>16199497
>destroys yourself
no hard feelings kids
Anonymous at Wed, 29 May 2024 20:04:35 UTC No. 16199536
>>16199497
This is sort of true but not exactly true.
Tabular methods are discrete set mappings. They map samples from one discrete set to another discrete set of some arbitrary but finite size through some abstract functional process.
Neural networks are not discrete set functions, they are smooth functions. In fact they are so smooth that they can be used to approximate any Lipschitz continuous function on the reals (up to some finite dimensional R^D) to an arbitrary degree of precision. This smoothness is actually what allows them to scale so well, it avoids the NP hard problem of an optimal tabular search (which is an integer programming problem and very computationally burdensome) and replaces it with (a generally larger but much faster) continuous floating point arithmetic problem.
Depending on the activation functions used, you could have more "saturation" like behaviors which would lead the NN to behave more like a tabular methods but this isn't the general way they function. In fact most of the important uses of Neural networks involve non-saturating functions like ReLu specifically because it doesn't saturate and as a result doesn't destroy as much information content from the data in the process.
None of this has to do with the "abstract reasoning" part of things, which I'd generally agree with the image in spirit. A Neural network itself is not doing any more "reasoning" than a Taylor series approximation of a function is. It's a function approximator, not a decision system.
You can build statistical decision systems by the connection of different network architectures, but I'd also argue that isn't really "reasoning" as much as it is inferencing via some "learned" optimal function approximation.
Anonymous at Wed, 29 May 2024 20:09:31 UTC No. 16199545
>>16199497
philososhit
"actual thinking" is a meaningless statement, whoever wrote this was a faggot
Anonymous at Wed, 29 May 2024 20:18:22 UTC No. 16199568
>>16199545
The neurological research into theory of mind is fairly interesting and still very much in a zone where there are many fundamental unknowns.
I'd agree that right now it is vague enough that most people are speaking more philosophically than scientifically/analytically, but there is still quite a lot left to "figure out" in terms of how organic creatures like humans process information and make decisions.
Anonymous at Wed, 29 May 2024 20:21:40 UTC No. 16199572
>>16199526
>>16199545
no one has an actual counter-argument to how OP's argument is actually incorrect. the main "counter-argument" is usually "your mom is an algorithm" or some dumb shit like that
Anonymous at Wed, 29 May 2024 20:24:41 UTC No. 16199577
>>16199572
I literally wrote one explaining the problems >>16199536
Neural networks produce smooth functions and as a result are not really comparable to tabular methods except in some "heuristic" sense.
That doesn't mean they are abstract reasoning, but his basic argument misunderstands mathematically how they function.
Anonymous at Wed, 29 May 2024 20:52:08 UTC No. 16199639
>>16199577
There are no smooth functions on a computer, it's all arithmetic. every function on the computer is approximated with a discrete approximation so "smooth" functions is only an approximate semantics. the real semantics is as it was outlined in OP's post, it's all integers and graphs of integer functions
Anonymous at Wed, 29 May 2024 21:01:54 UTC No. 16199659
>>16199497
Still they are more useful than you will ever be.
Anonymous at Wed, 29 May 2024 21:02:03 UTC No. 16199660
>>16199639
> the real semantics is as it was outlined in OP's post, it's all integers and graphs of integer functions
Yeah, except that it's not. When you implement the exp() function in whatever coding language you are using, it isn't performing a graph operation. You could try to conceptualize the function graphically but that's fundamentally not a meaningful way to interpret numerical analysis and will lead you to strange conclusions like "your computer has a table of values for this function" when its true behavior is not really anything like searching a graph.
Anonymous at Wed, 29 May 2024 21:31:31 UTC No. 16199717
>>16199572
>your mom is an algorithm
Correct
Anonymous at Wed, 29 May 2024 22:46:50 UTC No. 16199823
>>16199660
It is isomorphic. Don't be an idiot. You just want to obfuscate AI so that people don't reach the very basic conclusion: this shit is a stack of punch cards in a box, unconscious, unthinking. I don't know if you own nvidia stock or if you are a programmed NPC for someone who does.
Anonymous at Wed, 29 May 2024 23:04:46 UTC No. 16199854
>>16199497
You mean to tell me a series of deterministic logic gates aren't sentient? No fucking shit.
If it were true consciousness doesn't exist because it's deterministic. But consciousness and free will are de facto the ability to choose outside of cause and effect. To observe, assess, and then choose. The ability to choose must be fixed to something that has a desire for an outcome, and by our previous assertion it can't be deterministic it must be non deterministic therefore consciousness can only come from a non material soul not fully bound by the laws of this universe.
I fucking hate techniggers who want to puff up their egos and pretend they're god because they can fuckin sip mountain dew all night not getting laid and copy github.
Anonymous at Wed, 29 May 2024 23:14:10 UTC No. 16199873
>>16199823
Are you literally retarded? I'm not obfuscating anything. I don't think that the neural networks being smooth function approximators makes them any more "conscious" or "thinking" than any other numerical function you'd implement on a computer.
It's just not a graph/tabular search, and while in principle you could try and construct a graph search which on an "input-output" level behaves similarly, that's not how it works nor how it is implemented. You aren't doing yourself any favors by framing it as a graph search in an effort to convince people that NN's don't possess some magical thinking power, because they will rightfully dismiss you as someone who doesn't know how they work on a technical level.
Anonymous at Wed, 29 May 2024 23:24:52 UTC No. 16199890
>>16199497
Why do people feel the need to make this point so often? Like who is actually making the claim that AI is doing 'actual thinking'. Any time AI people get into that talk is just optimistic talk of potential future outcomes, the obvious implication being that these future AI system will be radically different to these current ones. It's like criticising a plane for not being a spaceship, just feels pretty retarded desu
Anonymous at Thu, 30 May 2024 01:02:02 UTC No. 16200031
>>16199890
New thing bad and the aifags that love to say the age of man is over. Every issue will spawn retards on each side.
Anonymous at Thu, 30 May 2024 01:31:36 UTC No. 16200060
>>16200031
>aifags that love to say the age of man is over
That's what I'm wondering, where actually are these people? I've watched an unhealthy amount of content where people talk about AI and the AIfag stance always seems to say that... some fundamental shift will happen with AI and only after this would we get AGI's or whatever. Something in the realm of 'actual thinking'. Maybe dimwits on Facebook or Twitter say the age of man is over in the immediate future, I've not seen it but I barely use that shit
I do have a few art friends and oh boy they use all the names in the book against AI stuff. The refuse to be anywhere near it or have anything to do with it. The level of seethe is astronomical, they get furious whenever I say AI is going to 'disrupt' sectors such as graphic design, but I'm just accurately describing how things are.
tldr; I think the majority of the retards are on the anti-AI side who seem to invent these imaginary AI people saying dumb shit, when in reality it doesn't seem to happen very often
webmind at Thu, 30 May 2024 01:33:10 UTC No. 16200061
>>16199854
Too much time is spent trying to replicate 'the mind'. Instead try to replicate 'life'. If you can get a machine to evolve first adapt to its environment, only then move towards intelligence.
Anonymous at Thu, 30 May 2024 01:35:08 UTC No. 16200067
>>16199497
This is literally Searle's Chinese room argument. If this is a CS person, they desperately need to read some philosophy.
Anonymous at Thu, 30 May 2024 01:47:20 UTC No. 16200092
>>16200060
I don't think you need "sentience" or '"real thinking" or really anything more advanced than what we have now for certain things to get bad.
As an example, most of the lower skilled mundane graphic design tasks that graphic designers do for their bread and butter (logos and graphics for local small businesses) already can be automated to some degree with current AI systems. None of the magical AGI stuff is necessary for this technology to already replace that one particular segment of the economy. While there definitely will still be higher skilled graphic design work that is done for more established customers I wouldn't be surprised if these kinds of "entry level" creative tasks pretty much completely cease to be economically viable soon.
Similarly, I don't think it really matters for chatbots if they are sentient. If they are "convincing enough" for some significant segment of the population to confuse them for interactions with a real human, they can be used to manipulate people in potentially dangerous ways. Again, none of the sci-fi AGI needed for that to be a concern.
Anonymous at Thu, 30 May 2024 01:56:43 UTC No. 16200100
>>16200092
This begs the question if some people are considered sentient. I'm talking the scizhos on each board, the <70iq types and even babies/little kids.
Anonymous at Thu, 30 May 2024 02:08:52 UTC No. 16200110
>>16199536
none of that shit matters since every neural network that you can actually implement in practice is a discrete set function
Anonymous at Thu, 30 May 2024 02:36:16 UTC No. 16200123
>>16200060
i mean, i have seen a few AI fags that are somewhat angry a the modern art "taped banana on a wall" shit from a subset of """"artists"""" and then somewhat project onto actual normal artist their gripes with the """"artists"""", which then leads them to go "fuck you" when the regular artist seems to be "gatekeeping" what art means, and since part of the AI fags grudge with """"artists"""" is the destruction/subjectivization/post-m
Anonymous at Thu, 30 May 2024 03:15:15 UTC No. 16200166
>>16200110
> Every neural network that you can actually implement in practice is a discrete set function.
I don't think you understand how floating point arithmetic works. You're mistaking the "input-output" relationship for the actual function itself.
Anonymous at Thu, 30 May 2024 03:22:06 UTC No. 16200173
>>16199497
AI doesn't exist. It's all a scam.
Anonymous at Thu, 30 May 2024 04:32:27 UTC No. 16200223
>>16199497
I have a feeling as AI progresses, we are going to be seeing a lot more midwit philosophy. It's not that philosophical discussion isn't warranted on the topic, but it's definitely strong retard-bait.
It's just another example in a long line where common parlance like "artificial intelligence" and "the computer is thinking" is shanghaied by some brainlet and spun into its most literal and obtuse form, so that it may be tackled by their philosophical mastery. The most common one as of late are those who hear the word "intelligence" in "AI" and then love to devolve into word salad about how it's "REALLY AKSHUALLY JUST A BAZILLION IF STATEMENTS SO IT'S NOT TRVE INTELLIGENCE!". They think their claim is profound, but it's just minimalization for the sake of philosophical masturbation. It's like saying digital computers aren't "truly digital" because voltage isn't fundamentally discrete, and that the digital illusion rises from our handling of a continuous signal. Yeah? So fucking what? I don't know if a word or phrase already exists for it, but it's purely "junk food insight" and complete fucking thought-slop.
Anonymous at Thu, 30 May 2024 05:07:39 UTC No. 16200242
>>16199497
That destroys every neural network even the one in your brain since it is just a fancy lookup table, you aren't actually thinking, your biological components are just referencing catalogued sensory information with a bit of randomness and degraded data thrown in.
Anonymous at Thu, 30 May 2024 06:14:09 UTC No. 16200341
>>16199854
>But consciousness and free will are de facto the ability to choose outside of cause and effect. To observe, assess, and then choose.
But choosing something because you observed it is cause and effect, its not outside of that at all.
Anonymous at Thu, 30 May 2024 06:24:59 UTC No. 16200365
>>16199890
>Like who is actually making the claim that AI is doing 'actual thinking'.
The people who try to claim that their consciousness makes them some kind of demigod, yet can't coherently define consciousness in a way that lets them retain their special status and exclude chatbots or roombas and have to conceded that those things do all the same things consciousness describes.
Anonymous at Thu, 30 May 2024 09:55:33 UTC No. 16200525
>>16200092
Good point in saying that it's still potentially dangerous and worthwhile of concern, even if it may not be AGI. That's like the adult/wise version of the anti-AI criticism, sadly this does seem to be rare especially when you talk to people 1 on 1. This "but it doesn't have a SOUL bro" argument comes up so often. Then I look around to find AI-minded people claiming it does... and can barely find anything. There's a deep insecurity there
>>16200123
Yeah postmodernism art arguments are cancer, can't stand them from either side. Another criticism I hear from art people is that AI isn't art because it doesn't draw shapes and shading like humans do, instead it's the denoiser statistical association stuff. Because it doesn't make art how humans do then it isn't 'art' in their eyes. Despite the end product having potentially more artistic value to the observer. I find that argument extremely weak, it's also a way that art people can dismiss AI art without explanation too. "The art wasn't made by a human, therefore it isn't art", just some ad-hoc belief made only to soothe the insecurity they feel
>>16200223
The funny thing to me is that philosophy could probably help AI development quite a lot. The problem being that most philosophy you see nowadays is just masturbation, they will craft some arbitrary barrier (specifically designed to reach the emotional conclusion they desire).
> I don't know if a word or phrase already exists for it, but it's purely "junk food insight" and complete fucking thought-slop.
I think the term is sophistry, sadly it's everywhere in modern day philosophy. Sometimes they will also use ultimate scepticism or anti-realism principles to disregard provable parts of reality, just because they don't like them. It's all very retarded and very exhausting
Anonymous at Thu, 30 May 2024 10:03:13 UTC No. 16200535
>>16200365
yeah well even their big main argument, the chinese room argument, that doesn't actually necessarily exclude how human consciousness works, as also said by this anon >>16200242
Anonymous at Thu, 30 May 2024 11:14:26 UTC No. 16200600
>>16200535
Don't you dare ever paraphrase (Me) to (Me) or my wife's son ever again.
Anonymous at Thu, 30 May 2024 11:59:50 UTC No. 16200646
>>16200600
You shouldn't have psychically pre-empted my flawless dunk on philosophers then. I will be contacting your wife in order to garnish your monthly MarioPoints stipend. I hope you learn from this. Be better
Anonymous at Thu, 30 May 2024 12:16:12 UTC No. 16200666
>>16199497
No shit. Any computer program is equivalent to some (discrete) mathematical function. Lmfao. What a fucking pseud. They teach you this in Maths For Gaytards 101 if you somehow manage to attend university. It doesn't really mean anything because there's no limit to what you can capture in a discrete function.
>>16200067
It's not... it's a more trivial observation than even le chinese room
>>16200242
You know "neural networks" aren't actually neurons right? Actual neurons are a lot more complicated. Whether they're somehow equivalent to a "neural" network is an unanswered question.
>>16200166
I don't think YOU understand how floating point arithmetic works. Do you think there are any infinite number of floats that a computer can represent?
Anonymous at Thu, 30 May 2024 12:18:43 UTC No. 16200667
>>16200365
Oh boy it's the Roombatard
Anonymous at Thu, 30 May 2024 12:25:37 UTC No. 16200674
>>16200365
The problem is that your definition of consciousness, which amounts to "a heckin roomba has it because it can vacuum around my table leg" is hopelessly inadequate. So your "criticisms" seem rather... retarded.
Anonymous at Thu, 30 May 2024 13:14:58 UTC No. 16200739
it's well accepted by people doing research in the field that current AI techniques (LLMS et al) are compression based, this approach was intentionally pursued because the training sets are so large for language that traditional techniques like tree search, a* pruning, backprop alone wouldn't work.
they use the language of "transformer" which is a subtype of sequence transduction modelling, which is what we would call data compression, the call encoder-decoder tokenization etc.
https://arxiv.org/abs/1706.03762
this was the seminal paper from 2017 that accelerated researched into these approaches and led to our current state of the art language bots. the same techniques also worked for image generation which is encouraging because it suggests there is a generalizable approach here, it also works for generating audio but the results there are less impressive, probably due to the difficulty of creating tokenized training data. there is a tipping point somewhere in the low billions (of tokens used - or using OPs analogy lookups possible) where you go from gibberish results to human like results.
the idea is that when you approach 10^12 (trillion) tokens you will have something like a conscious intelligent agent and if you keep adding to this you get AGI. the hunt is now on for ways of tokenizing the data faster and creating subsystems in the LLMs that do this, eg. human agents chatting and giving feedback, but ideally simulated human agents in their trillions doing this.
Anonymous at Thu, 30 May 2024 16:20:31 UTC No. 16200985
>>16200666
> I don't think YOU understand how floating point arithmetic works. Do you think there are any infinite number of floats that a computer can represent?
No, I don't think there are an infinite number of floats a computer can represent. What I do know is that your computer is generally performing bitwise arithmetic, not searching through some table of input-output mappings.
You can think of them as equivalent abstractions on paper, but in practice bitwise floating point arithmetic only needs enough memory to store the sequence of instructions and the sequence of manipulated numbers.
Thinking of this process as a graph is entirely misunderstanding how these functions work. There's neither "storage" nor "search" involved, which makes a very large difference in how the function actually works even if on paper you can abstract them as the same.
Anonymous at Thu, 30 May 2024 16:41:50 UTC No. 16201021
>>16199497
Did nobody tell that guy that all a computer can do is basic arithmetical operations? Calling them *just* or *only* arithmetical operations doesn't say anything except an arbitrary value judgment. Prove why the brain does something quintessentially different that couldn't ever be modeled by arithmetic.
All in all, the sooner you get over the coping and seething about how AI demeans humans, the sooner you can start thinking further. Unless you prefer the company of the same kind of people who were seething about Darwinism or Copernican astronomy because it hurt their fee-fees to be less of a special snowflake species.
Anonymous at Thu, 30 May 2024 18:31:50 UTC No. 16201156
>>16199545
The instantaneous impulses are force forward, there is nothing resembling thought or consciousness.
Anonymous at Thu, 30 May 2024 18:34:48 UTC No. 16201160
>>16200739
>consciousness is when big number
see >>16201156
Anonymous at Fri, 31 May 2024 05:55:19 UTC No. 16202101
>>16199497
Someone (karpathy?) also said the same thing about how "minimizing hallucination" is retarded because the hole thing only works because of hallucination.
If you get rid of hallucination it just becomes a lookup table
Anonymous at Fri, 31 May 2024 05:56:36 UTC No. 16202104
>>16199890
>who is actually making the claim that AI is doing 'actual thinking'.
Open ai grifters. AI saftey panel in government grifters. AI Doomsday grifters.
Anonymous at Fri, 31 May 2024 07:14:38 UTC No. 16202157
>>16200092
The current societal and economic paradigm is not exactly a utopia. Why is it bad if AI disrupts this?
Anonymous at Fri, 31 May 2024 09:20:29 UTC No. 16202233
>>16199497
If the point of the schizoidal faggot in the OP pic is discovering that any sentence in any language can indeed expressed as a very large integer, then yes that is correct.
What's the interesting thing about that discovery? Did we think we need some magic marmalade to be able to talk with each other and express meaning?
Anonymous at Fri, 31 May 2024 09:27:29 UTC No. 16202240
>>16200525
>Good point in saying that it's still potentially dangerous and worthwhile of concern, even if it may not be AGI. That's like the adult/wise version of the anti-AI criticism, sadly this does seem to be rare especially when you talk to people 1 on 1. This "but it doesn't have a SOUL bro" argument comes up so often. Then I look around to find AI-minded people claiming it does... and can barely find anything. There's a deep insecurity there
What? Consciousness? What the fuck has that to do with the AI problem space?
All of this is a complete (You) problem. You hang out in spaces that honestly baffle me. Why would I go on Twitter or Reddit to listen to their insane rhetoric how trans rights are human rights or how we need to establish world communism so they can be a useless fat slug?
This is the equivalent you are doing, reading takes by people who have absolutely nothing to add to the topic. Which ironically brings me to:
>>16200525
>The funny thing to me is that philosophy could probably help AI development quite a lot. The problem being that most philosophy you see nowadays is just masturbation, they will craft some arbitrary barrier (specifically designed to reach the emotional conclusion they desire).
How many modern philosophers have you read?
Do you know the writings of Mackie or Phillipa Foot? These are just an example.
You have not read one philosophy paper, so why would you give a strong opinion on such a topic?
Anonymous at Fri, 31 May 2024 09:45:49 UTC No. 16202250
>>16202240
>This is the equivalent you are doing, reading takes by people who have absolutely nothing to add to the topic
Fine, but that's 90% of the discourse on this topic. It's the argument laid out in the OP. It's essentially the Chinese Room argument. My point was merely that I find it strange why people time and again construct this 'counter' to AI, when I can barely find ANYONE making that claim in the first place.
It's like if your friend is constantly complaining that you say a tomato is a vegetable, they keep correcting this and bringing it up often... but you never even actually said that in the first place. It's ironically like these people are hallucinating things as much as AI does
I've read works of a bunch of philosophers, not sure what that would have to do with anything anyway... my point was that MODERN DAY discourse tends to follow the most junk ideas philosophy has shat out like ultimate scepticism or selective anti-realism, in order to further some ulterior goal. These principles are seeping into everything now via identity politics, you can see this philosophical junk seeping into biology when people say race isn't real, or that sex is a bimodal spectrum. Is all philosophy junk? of course not
Anonymous at Fri, 31 May 2024 10:16:52 UTC No. 16202274
>>16200061
I think the whole discussion of AGI should get us to approach consciousness in a different way altogether. It seems clear to me that the idea that brain is just a machine with inputs and outputs that somehow produces consciousness is false, since I don't see how AI is currently implemented (as in OP pic) will ever produce a sentient being. It may even be that "intelligence" as we understand it in the Cartesian sense doesn't even exist, it's just a peculiarity of Western humanist philosophy, so trying to "replicate" it in a machine to produce AGI will never work.
Over time I'm leaning closer to the idea that the entire philosophical background of Western civilization has us looking at consciousness the totally wrong way, that is why it will forever be a hard problem and will never be solved, nor replicated in a machine.
Anonymous at Fri, 31 May 2024 10:17:58 UTC No. 16202275
>>16202250
>My point was merely that I find it strange why people time and again construct this 'counter' to AI, when I can barely find ANYONE making that claim in the first place.
Stop being deliberatily obtuse. The popular cultural narrative with respect to robotics and AI for the last half century has been centered around robots and AI gaining sentience.
Anonymous at Fri, 31 May 2024 10:18:32 UTC No. 16202276
>>16202157
>picrel
Yeah, it's definitely the CO2 making things warmer, not the fact that 70% of the world's population lives in heat-absorbing asphalt jungles amongst millions of heat-emitting power lines, AC units, ICE vehicles, etc.
Anonymous at Fri, 31 May 2024 10:33:16 UTC No. 16202289
>>16199497
>bit flips exist
Don't you worry OP, we gonna have fun fine with our new AI overlords.
Anonymous at Fri, 31 May 2024 11:03:23 UTC No. 16202340
>>16202275
Yes you are right but when people make that claim it also involves a [magic step], from current LLM shit to something new. Or to be fair, sometimes they do say that with much higher compute and much more developed methods of LLM's, then we could find consciousness as an emergent property. These peoples arguments could still apply then, but in a world where LLM does lead to consciousness and these people are still arguing against it then their claim essentially boils down to robots not having a soul, which is just fucking retarded
No one (from what I can see) is claiming that these current LLM's are anything conscious. Criticism of LLM's for not being conscious therefore is dumb as fuck, and that's what happens constantly.
Anonymous at Fri, 31 May 2024 12:28:04 UTC No. 16202439
>>16202340
>No one (from what I can see) is claiming that these current LLM's are anything conscious. Criticism of LLM's for not being conscious therefore is dumb as fuck, and that's what happens constantly.
Word. It's like people forget what the "A" in "AI" stands for.
Anonymous at Fri, 31 May 2024 13:25:05 UTC No. 16202480
>>16200061
Evolution has no practical need to develop consciousness
Anonymous at Fri, 31 May 2024 15:14:44 UTC No. 16202590
>>16199717
Incorrect
Anonymous at Fri, 31 May 2024 15:16:50 UTC No. 16202593
>>16199873
OP explains in detail how it works on a technical level. It's just a bunch of numbers and graphs
Anonymous at Fri, 31 May 2024 15:22:00 UTC No. 16202597
>>16200242
False
Anonymous at Fri, 31 May 2024 15:25:36 UTC No. 16202605
>>16201021
Your mom is a number
Anonymous at Fri, 31 May 2024 15:42:51 UTC No. 16202634
>>16202605
Your mom is a stupid cunt
Anonymous at Fri, 31 May 2024 15:44:32 UTC No. 16202638
>>16202593
You could make a similarly reductionist argument about the brain, it's just a bunch of neural cells with ion concentration gradients, action potentials, neurotransmitters, etc. Yet clearly highly complex behavior emerges out of relatively simple parts.
Anonymous at Fri, 31 May 2024 16:23:09 UTC No. 16202678
>>16202638
Invalid comparison
>>16202634
She's not a number tho