Artificial Intelligence Isn't Real

preview_player
Показать описание

This video has been approved by John Xina and the Chinese Communist Party.

Attribution (email me if your work isn't on the list):
Рекомендации по теме
Комментарии
Автор

Thanks for tuning in to today's video!

AdamSomething
Автор

As a programmer that studies AI, we almost never actually use the term artificial intelligence, we usually just say machine learning, as this more accurately describes what is happening.

SurfingZerg
Автор

I'm opposite of *Artificial Inteligence*, because I'm *Natural* and *Stupid* .

mateuszbanaszak
Автор

Personally what pisses me off is when someone takes one of those AI chat bots, and gives it some random character from any show, and goes "Omg this is basically the character" and it just gives the most basic ass responses to any question

justpassingby
Автор

I've done a PhD in machine learning, I design machine learning system as a job
* Yes, AI is a very poorly defined word, which have been stripped of the little meaning it might had because how much it was stretched in all directions
* Intelligence is not a boolean feature, it's a continuum. Where do we put a virus ? The most simple unicellular organisms ? Industrial control systems are on the level of a simple bacteria in term of complexity and abilities, minus the self-replication ability (3d printers are this close to cross that gap)
* Your cat example is a very good explanation of what statistical inference.
* You can implement statistical inference in various ways, one of which is neural network. Neuron network can have internal models that does what you call "the intelligent way". That internal model is not set by the programmer, it's built by accumulating training on randomly picked examples aka stochastic gradient descent.
* The Chinese Room argument have its critics, some of which are really interesting

And yes, there are tons of cringe bullshit on this topic, to the point I carefully avoid mentionning I do AI, I say I do statistical modeling

alexandredevert
Автор

"Behaving perfectly like a human doesn't mean they are intelligent" is a sentence that can be used on quite a lot of people too lol

nisqhog
Автор

artifical intelligence is when the computer contolled trainers in pokemon use a set up move instead of attacking

flute
Автор

The only fear I have for current "AI" is companies betting too hard on it, and having it destroy them. Not by some high tech high intelligence AI takeover, but by the AI being poorly implemented and just immediately screwing over the company, like "hallucinating" and setting all company salaries to 5 billion dollars.

Cptn.Viridian
Автор

If people insist on using "AI, " I propose we call it "Algorithmic Intelligence" because that's far closer to what it really is than Artificial Intelligence

mistgate
Автор

Robotics/AI researcher here, you're definitely right to suggest that AI is being completely blown out of proportion by the media. That being said, certain things you mentioned computers not being capable of they certainly can do, it's just a case of that they're currently still the kinda things that are being developed in research institutions and therefore not viewable by most people. For example, the fat cat example could be tackled by a combination of causality and semantic modelling could represent the relationships between feeding the cat and its weight. Furthermore empathy modelling is also an idea within reward-based agents/robots, effectively having the robot reason about whether an outcome would be optimal from the perspective of another being (e.g. a cat). Of course we're still a long ways off, but that is more of a software/theory issue than a hardware issue, in a sense, we have all of the machinery we need to make it happen, it's just a matter of knowing how to structure the inner workings of the AI that's the difficulty.

With regards to the Chinese room thought experiment, it's worth mentioning that only one school of thought precludes this disproving consciousness. I'm fairly certain that if a baby could talk, and you were to ask it whether it understood anything it was experiencing, I doubt it would, yet I don't think anyone is arguing that babies are not conscious. Even that aside, I think what ultimately sets aside human intelligence, and what will ultimately set aside future AI, is the ability to reason about reasoning, or in other words meta-reasoning. This is currently quite difficult considering the biggest fads in research right now involve throwing a neural network at problems, effectively creating an incomprehensible black box, but there's definitely the baby steps there of making this happen.

All that being said, I totally get why you made this. The way everyone's talking these days you'd be forgiven for thinking the machine revolution is due next Tuesday.

rhyshoward
Автор

Incredibly brave of Adam to stuff his cat with food to the point of morbidty obestiy just to prove the limits of AI, that's real dedication.

Movel
Автор

I'm a data engineer for a company you've heard of. I fully agree that the general public doesn't really understand or properly use the terms "AI" and "Machine Learning". However, I would argue that in so many cases neither do the "tech people".

kcapkcans
Автор

I once passed a basic french speaking exam with essentially no comprehension of what I was saying. Just copied the tense structure of the question, added a few stock phrases and conjuctions, and sprinkled in some random nouns and adjectives that I couldn't for the life of me tell you what they meant, only that my brain somehow decided they were in the same topic. They were testing for compression; I used a different method to have the appearance of it.
AIs work with similar logic, doesn't matter how you get the results within the task, so long as the results appear correct.

aliceinwonderland
Автор

I have an MSc in Computer Science and Artificial Intelligence and I can say that how we use these terms and how the media uses these terms are very different. "AI" is a huge field that refers to basically anything that a computer does that's vaguely complex. So when your map app tells you the shortest path from A to B, that's AI, specifically a pathfinding algorithm. We we talk about stuff like chat gpt, we wouldn't really call it AI, because AI is such a general term. It's Machine Learning, more speficially Deep Learning, more specifically a Large Language Model (LLM). Stable diffusion is also Deep Learning, but it's a diffusion model.

stevenstevenson
Автор

I disagree: people are scared by AI not because they think they're seemingly "human", but because perfectly acting like one without understanding the meaning behind it can lead (in the future) to real life consequences.
If you teach an AI to hack your computer and delete all your data it doesn't matter if it understands what it's doing as long as the action is being done.
Not having free will doesn't mean not creating consequences; if anything, it's worse.

ItaloPolacchi
Автор

As a programmer, thank you. It's annoying to have to explain to people that AI is not intelligent, it's just an advanced data sorting algorithm at the very most. It has no thoughts, it has no biases, it has no emotions. It's just a bunch of data sorted by relevance. This isn't to downplay the technology, the technology behind it is stunning, and it has good applications but to call it intelligence when it isn't is absurd.

thrackerzod
Автор

Thanks for the video. My main worries over this are that executives believe that this can replace human workers, apply this "AI" to everything fire a lot of people and then they end working the remaing ones to death when those "AIs" fail in doing their tasks because nobody would know if its outputs, and/or inputs, are accurate enough; or if, the heavens forbid, they gave IAs any type of decission power.

miasuarez
Автор

As a CS Student I agree with the main point of the video, but I'll just throw it in the room that we actually don't know what "real intelligence" really is. So maybe at some point AI will actually become "real" without any way to tell it apart. We just don't know.

RoiEXLab
Автор

The thing I'm afraid of is that corporations couldn't care less, as long as they don't have to pay actual humans for being artists, programmers, etc etc, they will be using AI even if everyone knows it actually has no idea what art or code is

bettercalldelta
Автор

The cat should make more appearances in your videos

stevejames