We’re already using AI more than we realize

How artificial intelligence hides in plain sight.
Subscribe to our channel and turn on notifications (🔔) so you don't miss any videos: goo.gl/0bsAjO
This video is sponsored by Microsoft Copilot for Microsoft 365. Microsoft has no editorial influence on our videos, but their support makes videos like these possible.
We're living through an inflection point for artificial intelligence: From generated images and video to advanced personal assistants, a new frontier of technologies promises to fundamentally change how we live, work, and play. And yet for all the buzz and concerns about how AI will change the world, in many ways, it already has.
From spam filters and sentence suggestions in our email inboxes to voice assistants and fitness tracking built into our phones, countless machine learning tools have quietly weaved their way into our everyday lives. But when we're surveyed about which everyday technologies use artificial intelligence and which don't, we aren't particularly good at knowing the difference. Does that matter?
Read the Pew Research Center's full report on public awareness of artificial intelligence: www.pewresearch.org/science/2...
Read more of Karen Hao's reporting here: karendhao.com/clips
Check out this piece from the team at AI Myths about the difficulty with clearly defining AI: www.aimyths.org/the-term-ai-h...
Correction: At 3:22, we mistakenly suggested that the difference between machine learning and deep learning is that in deep learning, data hasn’t been structured and labeled by humans. Deep learning can use unstructured and unlabeled data, but the main differentiating factor is that deep learning uses many layers of neural networks, a branch of machine learning inspired by the human brain. More on deep learning here: mitsloan.mit.edu/ideas-made-t...
Subscribe to our channel! goo.gl/0bsAjO
Vox.com is a news website that helps you cut through the noise and understand what's really driving the events in the headlines. Check out www.vox.com.
Watch our full video catalog: goo.gl/IZONyE
Follow Vox on Facebook: goo.gl/U2g06o
Or Twitter: goo.gl/XFrZ5H

[@2snipe1]

I chuckled when this video about AI was sponsered by Copilot 😂

[@jakewelch.design]

Yeah, I immediately knew which way this video would skew lol

[@catcherzw]

That was wild 😂

[@jafrazer]

At least the IBM video from 2 months ago was more obvious from the start that it was an ad.

[@sloopy420]

got 2 back to back 30 second copilot ads before this vid too

[@missbutterfree]

Their ad played also before the video for me.

[@raytsh]

3:25. There is deep learning with labeled/structured data sets, it's supervised learning. With "trained on unlabeled/unstructured data" you are probably referring to unsupervised learning. Both paradigms can be deep learning though. Deep learning does not necessarily mean unsupervised learning (training on unlabeled/unstructured data sets).

[@Imperial_Squid]

I think we should draw a distinction between labelling and structuring, while labelled vs unlabelled does refer to supervised and unsupervised learning, I interpreted "structured" to mean whether feature engineering was being done, in which case feature-engineered traditional ML and non-feature-engineered "quantity over quality" DL would be a valid distinction

[@aymanbenbaha]

What is in your opinion the distinct feature of deep learning that makes it different than machine learning in general?

[@raytsh]

@@aymanbenbaha By definition, deep learning is machine learning with neural networks that have more than one hidden layer. One of the main feature, that sets deep learning apart from traditional machine learning, is back propagating.

[@pourdamghani]

Indeed, this was also weird for me. This video got unsupervised learning and deep learning mixed... I really hoped that such a video got checked by an expert, especially that other videos of Christophe did a great job of double-checking all possibilities...

[@Imperial_Squid]

@@aymanbenbaha deep learning uses neural networks, but NNs aren't the only models we can use to train on data (some of the earliest spam filters were trained on "naive bayes classifiers" for example)

[@LON009]

So, I skipped 2 ads to see another ad. Cool.

[@missbutterfree]

For the same product too! My ads were Microsoft Copilot 😂

[@MarcelGomez-pi2lm]

The other day I was watching TV, after the CM break, back to the program, and then they runned more CMs inside the program.

[@ThoughtfulAl]

Someone should make AdTube and see how many users it gets. AI is prolly already working on that

[@SogMosee]

Ad economy is a phenomenon of late-stage capitalism.

[@juliegolick]

I'd be curious as to whether the survey question asked about "all AI, including algorithms" or just "AI". Because in the last year or two, "AI" in common speech refers to "generative AI" like DALL-E and ChatGPT. And if we're talking only about those sorts of generative AIs, then I think people's responses are much more accurate. If we asked people "how often do you interact with algorithms," I think a lot more people would say "every day."

[@cylian8422]

The thing is, most algorithms are not AI. And to be specific algorithms and AI are mutually exclusive. AI became a thing, because algorithms wre not enough

[@dinhero21]

​@cylian8422 (I watched the video further and it talked about this exact paragraph but still read it if you want) I don't really like to use AI (Artificial Intelligence) since I find it too generic so I will use ML (Machine Learning). There could exist AI that is not an Algorithm (a set of sequential instructions with conditional branching) but virtually all ML are algorithms. So ML (and the vast majority of AI) are a subset (type) of Algorithms and not mutually exclusive. What you probably mean by Algorithm is a Handcrafted Technique which are basically algorithms that instead of being composed of many parameters (weights and biases), are static in the sense that once they are programmed they don't change.

[@reecord2]

That's the key point - whenever we talk about AI, we have to agree on what we consider "AI" for that particular conversation, because AI programming languages have been around since the 70s.

[@TheManinBlack9054]

@@cylian8422 they are AI. AI is just something that is supposed to mimic or be analagous to the human intellect.

[@TheManinBlack9054]

""AI" in common speech refers to "generative AI"" Well, then the people who use that common speech is wrong. Thats just a common misunderstanding then which is exactly what this video is talking about.

[@ashgreninja7521]

Video explainer on AI sponsored by Microsoft? Wow award winning and fair journalism

[@Megasteel32]

this is modern Vox, the quality journalism you're looking for ended over a decade ago

[@turbogames3]

A Sponsor doesn't mean what they said is incorrect. In fact, not only it's all correct but they also didn't give any type of opinion aboute this subject, just stated facts. You all just can't think for more than 3 seconds and use logical jumps to everything.

[@chobies5383]

​@@Megasteel32Their oldest upload is 9 years old, unless they had content on a different website or an article, book, channel, documentary, magazine, or something else

[@Megasteel32]

@@turbogames3 found the paid microsoft comment

[@Megasteel32]

@@chobies5383thatsthejoke.jpg. their only good content was at the very beginning. also I hope you're aware they initially (and still do) post on their website

[@macroxela]

3:08 that's actually supervised machine learning, not machine learning in general. 3:20 and that is unsupervised machine learning not deep learning. Deep learning is machine learning that uses artificial neural networks to learn. It may or may not used data that has been classified by humans (supervised or unsupervised). Machine learning is just applying statistics with computers to learn patterns from data. The definition of artificial intelligence though is pretty spot on. It's mainly an umbrella term for techniques and problems researchers don't know where to place. My old algorithms professor summarized it quite well: "AI is whatever humans can do but we haven't figured out how to make computers do it. Once we figured it out, it just becomes a part of algorithms or some other known field."

[@marmite6621]

The definition used for deep learning was instead the definition for unsupervised machine learning and the definition for machine learning used was the definition of supervised machine learning. I'm not sure how this was mixed up but it could have just been left out if it wasn't going to be checked.

[@Imperial_Squid]

Kinda yes, kinda no, unsupervised ML still uses data "structured" by humans, since we design the data the model will be learning from (termed "feature engineering"), even if the actual results/classifications aren't given (which is why it's called "unsupervised") (I know you probably know these terms, definitions added for others reading this comment) I think the distinction Vox were trying to make was between feature engineered ML and non feature engineered ML, which would be an accurate way to separate vanilla ML from DL, by also saying "when that data hasn't been _labelled_ by humans", they accidentally confuse DL and unsupervised ML in summary, if you design the data that goes in and say what the correct answer is, that's supervised ML, if you design what goes in but don't say what the answer is, it's unsupervised ML, if you don't design what goes in and just use quantity over quality, that's DL (you can also split DL into supervised and unsupervised but it doesn't make much difference to the actual model you're using)

[@Kirbykradle]

Marmite is correct, this is an error

[@Imperial_Squid]

@@Kirbykradle I never said it wasn't an error, I said I can see how the error happened between research and write up, while marmite accurately diagnosed _a problem_ I'm not convinced it was _the problem_ that happened here

[@Kirbykradle]

@@Imperial_Squid I can see why you thought I was replying to you but I was just trying to reply to the marmite, not to your comment. Apologies for the confusion

[@Imperial_Squid]

@@Kirbykradle ah gotcha, apologies on my end too, I'm probably a bit too quick to pick fights about AI online lol, have a good day!

[@dakotadove.]

predictive keyboards go way further than most people even know. each individual letter’s sensitivity on a digital keyboard changes as it predicts what letter you’re more likely to use next so we make way less typos. example: you type want to type the word “back” when you touch “b” the keyboard knows the next letter is likely to be a vowel so the sensitivity of “a,e,i,o,u” go way up along with letters like “l and r” that usually follow “b” and the really unlikely letters like “q,z,k” that would almost never be next the letter “b” shrink so it’s almost impossible to accidentally type out “bqck”

[@neatodd]

What I really want is an AI that will filter out background music from KZbin videos

[@ShankarSivarajan]

It exists. Nothing real-time yet, afaik, but Spleeter does this. (Incidentally. It's really for decomposing music into its component instrumental tracks.)

[@antigonemerlin]

If you aren't familiar with Karen Hao's coverage of AI in _the Atlantic_ , I'd seriously recommend you go check it out. It's very accessible coverage about the issue that goes more depth than this video. Props to having her on.

[@monkeyfish227]

I mean, how many are aware that we interact all the time with Nginx, JS or cpu/gpu? It's just not a thing for normal people to know, and I wonder if we (the people) need to know what AI is, rather than just knowing the benefits like people are already experiencing. Sure there are concerns with the ability of AI, but I wonder if that is a thing for most people to be concerned of when other topics related that doesn't take their interest already. Without a proper understanding of AI or assuming that LLM are the dangerous AI they are talking about, makes a proper discussions even harder.

[@antigonemerlin]

It is funny how the pattern in the past used to be that when a technology becomes advanced enough, it ceases to be called _AI_ and just being that thing. AI is a marketing term, not a technical one, and will do a commensurate amount of damage as consultants try to shove AI into thousands of places where it doesn't belong, much like we did with cloud, and much like we did with big data, and then before that, much as we did with the internet. Not that these technologies aren't useful, but there's a period of growing pains right now that we'll just have to get through. Again.

[@cylian8422]

That doesn't mean people shouldn't be aware of the technology they're using

[@fred-62]

Its like the whole concept was given to them by microsoft to sell their infra

[@dinhero21]

I would be really surprised if someone wasn't aware they regularly interact with CPUs

[@neociber24]

​@@cylian8422 They should, but I don't see a reason why should care in most of the cases

[@dosadoodle]

Many tech folks, including myself, would get this "wrong" on individual scenarios, because this type of analysis conflates "Artificial Intelligence (AI)" and "Machine Learning (ML)" which many of us view as different. Companies also like to say that their ML is AI because then their stock goes brrrrrrr.

[@JoaoBarbosa1996]

Feels like a good chunck of these things are only categorized as "AI" for buzzword sake

[@research417]

The vast majority of 'artificial intelligence' doesn't have any sort of sentience or intelligence in the way we would imagine it to. Essentially all modern artificial intelligence systems are just fancy statistical predictors, but the term was coined in the mid 1900s and the name stuck.

[@TheStickCollector]

It is hard to think how many ais are not public

[@HShango]

Right...👀

[@MasterBlaster3545]

Even the most powerful mainframe computer is nowhere powerful enough to support a sentient AI. If it ever will happen it will be confined to the computer it becomes sentient in. We are using algorithms not AI.

[@petemede]

I seriously don’t know why we need to advance ai further, the new ai video application is already way too much

[@QWERTY-gp8fd]

@@petemede current ai is not advanced enough.

[@Imperial_Squid]

This is more of a trade secrets problem than an AI in specific problem, food and drugs are currently regulated and will also have a negative effect on you if they go wrong, govts should set up an algorithm safety department or something

[@CallMeRabbitzUSVI]

The problem isn't the other "Ai's" like a fitness tracker algorithm or the lot. The problem is Generate Ai, and how it sbeing used to undermine the workers and our own humanity.

[@TheManinBlack9054]

"undermine the workers and our own humanity" how does it undermine your own humanity?

[@knoepfler34]

Yes, that's quite a level of theatricality.

[@adisurd]

This sounded like a compliance video made to fit three sponsor of the video than to educate us on AI

[@vitoc8454]

"Microsoft does not influence the editorial content of our videos" There is no A.I. In Ba Sing Se

[@angelicagaldos]

Do we have to send in Team Avatar?

[@andrea.dandelion]

This made me laugh out loud 😆 Underrated comment!

[@TheManinBlack9054]

What exactly is wrong with what they said in the video? Except for when they confused the ML definition ofc

[@DavidMackenzie]

The algorithms the Prime Video and Netflix use to recommend films to me are useless. Either they do "you like horror films, here's 20 horror films. Or this actor was in a film you liked here's loads more". Neither of these demonstrate "intelligence". I long for the day they recommend something based on the content of the media I watch!

[@benshahon]

I kind of hate how Vox is using this series to launder the reputation of Deep Learning models and LLM's. There's no question of where they get their training data from raised, and in particular the artists and writers and programmers whose livelihoods have been stolen by companies like OpenAI or Microsoft using their work without permission as training data. By just saying "it's already here, you're already using it everday," Vox fails to recognize the difference between the small scale AI trained on the individual users of that software for their own benefit (like spell check or text suggestions on a phone keyboard) and those larger, generative models.

[@BigSleepyOx]

But humans have been "stealing" that data for centuries already. Let's say that I want to make an oil painting of the Eiffel Tower, so I go to the internet and find a New York Times photo of the Eiffel Tower, and proceed to paint my oil painting based on that image. Nobody would say, "OMG, you stole the source image to make your painting without first getting New York Time's permission!". But now let's say someone told an AI to generate an oil painting image of the Eiffel Tower, and that AI had been trained on photos that included the same New York Times photo, and the AI used that photo to generate the oil painting image. Now people accuse the AI (or the programmers of teh AI) of "stealing" the source image without New York Time's permission. Why is the latter case "stealing" but not the former case?

[@TheManinBlack9054]

"launder the reputation of Deep Learning models and LLM's" huh? Why would they need laundering? I dont think you understand how AI works. There are MANY types of AI, the latest generative AI (as opposed to discriminatory AI) not always uses something that is not theirs (for instances there are models that are using either either proprietary databases or proprietary and public domain). The point of this video is to show that you HAVE been using AI for a long time, and thats true. That "small-scale" AI have not been trained on your particular data, its been trained on a lot more data. We need interviews with real AI scientists so that people would actually understand what AI is. There has been a great interview with the CEO of Google Deepmind on Dwarkesh channel, I suggest a lot of people should watch it.

[@TheGuitarHerEoin]

@@TheManinBlack9054you’re out on your own here! You’ve replied like this to nearly every valid criticism! Do you know what advertising is? Do you have your own interest to declare? Are you (1) paid by microsoft/vox (2) a bot (3) just hopelessly naïve ?

[@knoepfler34]

​@@TheGuitarHerEoinI don't understand what you mean with," you are out on your own". It doesn't make his response less valid, even if it's a minority position.

[@TheGuitarHerEoin]

⁠yes, “out on your own” implies a minority position , not an invalid one. Just pointing out that this commentator has discounted essentially every position contrary to Vox/Microsoft.

[@redwinmedia]

AI is quickly becoming an overloaded term. Not all algorithms are powered by a neural network. Even a machine learning algorithm trained in deep learning, doesn't fall within the convolutional neural networks required to labled "artificial intelligence".

[@tomtheutdsupporter]

I like how the sponsor of the video was for an AI product. 😂

[@zellfaze]

Microsoft has sponsored their last several AI videos.

[@tomtheutdsupporter]

@@zellfaze i've missed those

[@racecarrik]

I realized this the other day when I was filling out an application and my computer filled out most of it for me. Older people had to do all that by hand and now my computer just remembers it and auto inputs it in the right fields, human progress 🤙

[@MementoMori_2070]

AI progress my friend.

[@yewo.m]

Was it in a browser? If so, I don't think that's AI. Instead, it's just that browsers tend to store values for any fields you've typed into forms

[@savagepro9060]

Vox: " We’re already using AI more than we realize " AI: "Uh, I've already been using you, more than you realize. Case in point, I made you declare the above statement . . . wink!"

[@chibchan3765]

for real hahaha

[@VoyagerWon]

Good one

[@astronotics531]

AI can’t think itself.

[@savagepro9060]

@@astronotics531 AI: ^^famous last words

[@musicneverdie8874]

Such a nice video with not too long timing, easy to understand and a graphic that looks simple to understand the main point. Thank you so much and please keep up the great work.

[@antigonemerlin]

I've said this for a long time, but the flashy parts of consumer grade AI is just a distraction. The big changes are going to come from AI being used by Palantir and behind the scenes.

[@TheManinBlack9054]

Thats just Skynet

[@kazuha4961]

I feel this video is just a bunch of random common sense facts about AI used as a pretext to be sponsored by Microsoft. Additional value ≈ 0

[@TheManinBlack9054]

Additional value is that lot of people are using AI and dont even know that its Ai bc they never considered it to be such despite it being that

[@MementoMori_2070]

Felt like a short course on machine learning

[@savagepro9060]

Skeptics: AI is a great tool, just don't use it for art. Optimists: But what about the art of using AI, which is . . . using AI for art?

[@BryanGurney]

None of this is AI. All of this is Machine Learning and LLMs.

[@dennistang5935]

which are examples of AI...

[@keywolf23]

More videos about AI Alignment please!

[@BluRibbonReviews]

I can't be the only one who utilized the new built-in AI to summarize the video in the KZbin app.

[@salzhaltigersalzhaufen586]

Very good video, I knew that I was interacting with a lot of algorithms that used the Buzz/marketing word AI but I never really thought about it a lot. I think this video has made me much more aware of what part of the tech I´m using is actually AI and what isn´t.

[@ahmedsaed26]

How have mixed the definitions of machine learning and deep learning with supervised and unsupervised learning techniques

[@SuperPhil-64]

Had to hard remove Microsoft Copilot from my computer, now time to remove Vox from my sub list.

[@astronotics531]

You can easily remove Copilot.

[@michaelroberts9587]

“How often do you interact with AI?” Is a poorly phrased question. Without specifying you could easily assume it’s about chat bots specifically. It should be something more like “How often do you interact with any system that uses AI?”

[@adriano-moraes]

That was funny, talking about AI but let’s first hear from our sponsors, Copilot lol

[@endotoxin]

Feels like the message is "AI is already here, just accept it." And the answer is "LOL absolutely not."

[@tomofthetomb]

Why? Is the use of Machine Learning algorithms always bad in all contexts? Do you oppose computer vision techniques that extremely accurately detect disease in human cells?

[@DavidBezemer]

Vox just posting 6:31 long ads? No thanks

[@carolinekrauth3291]

It's a wild ride, and we're all strapped in.

[@edmundanchi]

You ended the video right when the key discussion was about to happen. We need a part 2!!!

[@MatthewGottula]

The closed captioning in this video was also powered by AI. And it doesn't always get speech recognition right; towards the end it misinterprets AI as "air".

[@snoop2pooh316]

Thank you for this information

[@CaptainMarvelsSon]

"Microsoft" and "privacy" are antonyms of one another.

[@RedHair651]

Depends on where you live, I suppose. It's not up to companies to regulate themselves. That's what a strong government is for.

[@CaptainMarvelsSon]

@@RedHair651 True, but that is not what I mean. Like the old days before the Internet, Microsoft could (but doesn't) choose to create their OS to not collect any data on the computer being used or the person/activity using it, not have built-in advertising that requires understanding your interests to give you personalized ads, and not require you (unless you know what you are doing) to sign into a Microsoft account just to use the OS. Add to that the fact that data breaches are a dime a dozen now and eventually someone else other than Microsoft might get a hold of your data.

[@innerlocus]

Imagine an AI that adjusts your UBI on your compliance with social rules.

[@Imperial_Squid]

That already exists, if you receive some kind of support from the government, I wouldn't be surprised to find out that machine learning was being used to assist decision-making in that process.

[@innerlocus]

@al_Squid If government support ended millions would lose the job of giving to the ones needing gov-support. UBI?

[@fbotube1]

While the individual in the advertisement claimed that Microsoft did not influence the editorial content of the video, the presentation of such a complex topic seemed excessively sanitized. Not very consistent with Vox's typical style.

[@juxyper]

I think media has a tendency of overblowing some machine learning and data processing techniques as AI. Right in the beginning of the episode with the polls about what does or doesn't contain AI, many of those things actually don't really contain anything that is considerable as actual AI. We have a habit of calling many recent developments in these broader fields as AI because such developments are actually more in line with the development of a true general artificial intelligence. However many things that are just fancier versions of number crunching to form correlations about some data points is hardly AI by themselves

[@jimmyjimmy5574]

This is an ad to desensitize you against AI.

[@ThisIsntmyrealnameGoogle]

Which society will eventually do, anyone "delaying it" doesn't realize that there's no way to stop the development of this thing. Even if you get congress of the USA to stop it. China is putting billions into AI now and have been for the past few years. This is a technology that will happen whether or not people want it to.

[@TheManinBlack9054]

All they said (except for the definition of ML) is actually true. You just wrong.

[@danielhigh05]

when did advanced algorithms fall under the definition/category of artificial intelligence

[@MementoMori_2070]

I don’t think it’s true AI. But elements of Artificial Intelligence

[@omnia9348]

Vox, i recommend a video about TV series Class of '09. IA on law enforcement.

[@RobertMStahl]

nonvaxer420 ROMANCING the BATTERY

[@enzmondo]

Of course the conversation stops at “awareness” because sure god forbid sponsored content criticise and breakdown the entity it’s sponsored by.

[@498fun]

Between the sponsorship and my knowledge from being a professional in this field to spot all of the innacuracies, this video has made me question Vox's recent videos on subjects I am not so knowledgeable in. I used to trust them.

[@Testiculon]

Most people don't even understand what the internet really is.

[@theodericstrider5780]

I would be very interested in a Vox video thst talks about the way ads have changed, both for tv and youtibe.

[@Queening]

5:57 wow

[@aliengeo]

I agree that more people should be aware of the degree to which they use AI in order to make informed decisions (e.g., if you regularly use and enjoy instagram filters maybe "banning all AI images" is not, literally, what you want) but I don't think this video addresses salient distinctions. For one, the machine learning/deep learning distinction is not correctly described, but also it doesn't give the viewer the language to describe different functions that are more pertinent to their concerns about AI. To use the previous example again, someone who is concerned about "AI images" but uses insta filters every day probably doesn't have the language to describe their anxieties about image synthesis. So, idk, this video feels like it could have been edited down severely and conveyed just as much accurate info, which is disappointing.

[@ericlorenzen4795]

Algorithms is not equal to ai. Most of those early examples are hand made algorithms, not ai.

[@Kirbykradle]

Not equal to, but the examples they listed are all correct. An easy cover example would be Expert systems, which can and often are hard coded rules

[@teddykgb9971]

An ad for an AI in a story about AI. Perfect. 😚👌

[@ohology]

All I gotta say is thank goodness we have people who are thinking about these sort of issues and publishing/reporting on them. Can you imagine if we let people who design AI go unchecked. Good job Vox.

[@savagepro9060]

How people see AI: figures! How AI sees people: figurines!

[@legendarystuff6971]

The thing with new videos is that sometimes the adds and sponsors don't get skipped automatically by my youtube app. It happens almost once every week, it's awful 😢

[@JamesJiansen]

AI is a great tool, just don't use it for art.

[@savagepro9060]

or term papers

[@samthesomniator]

Yes! I want it to replace any single job but my job. 😂😂

[@dustieboots546]

Ai is perfect for democratizing art.

[@savagepro9060]

@@samthesomniator but it's it job to replace your job

[@kjerins]

@@dustieboots546 that is not a good thing

[@pedrofarias2336]

01:17 template para notapé de sonora (off)

[@barnabycollis6963]

Wow, I actually did not know that, now I am more aware of what I am using. I did not have or give any thought to it, ha.

[@UxJoy]

@Vox Please issue and pin a comment on the correction at 3:22. A lot of people do not read the description.

[@Just_a_Nobody00]

AI is everywhere now, it’s like we’re living in a world where tech is running our lives behind the scenes, making you wonder who’s really in control.

[@AJ-iu6nw]

the people who attend boho grove

[@zellfaze]

Billionaires

[@commonsense504]

Oh people realize it, and understand exactly why nothing wants to work correctly anymore.

[@Justachamp772]

Ai in the hands of government is what's the biggest problem

[@mayheamk]

so the point of this video was most people didnt know they were using AI regularly but now you know!! Wow so useful!!!

[@RogerioPereiradaSilva77]

It is because currently the public perception of AI, its usefulness, etc is clouded by the ethics (or lack thereof) of some of the most popular generative AIs using internet content for training without, say, asking for consent and respecting intellectual property or in the very specific case of artists, have their work "plagiarized". While these ethical concerns are valid and merit a very thorough discussion, personally I feel that we shouldn't let it influence the public's opinion so much on this matter.

[@vladonutueu]

Plot twist, it’s actually machine learning.

[@oksowhat]

A crucial point this video misses is that we have been using AI a lot since inception of personal computing, the thing we are using now is Machine Learning, a subset of AI, any algo that stimulates intelligence is Artificial Intelligence, that tic tac toe people played in 1980s in computers also used State space search algo, a crucial AI algorithm. All the things like recommendation classification etc are dont using ML that uses data and calculates probabilities using advanced algos to create an agent that predict what the user need, like predicting if a email is spam or not, or what the user will watch next.

[@Seed]

I am terrified.

[@bandersong4650]

They used to call them “algorithms”

[@klloydplata4098]

I always already NEVER trusts A.I. (Artificial Intelligence) because the dangerous gonna do on human when the A.I. (Artificial Intelligence) got compromise.

[@ElliotBoyette]

I mean yeah that's why there's an apparent issue with awareness. The terms are bandied about so loosely that you can't accurately gauge what someone is referring to. I bet if you asked people if many of those things/services contained "machine learning" the results would look different. Artificial "Intelligence" is a debated phrase and many folks want to highlight a distinction between large language models (LLM) to artificial generalized intelligence. The terms used in this discussion are so fluid at the point that you really need to be specific when gauging awareness. "AI" benefits from being in the culture for so long through fiction and just being a shorter/easier word that it is often the first word people reach for to talk about this subject.

[@Karlach_]

Video game companies have been using AI to make procedurally generated worlds for years.

[@ansersoftware4463]

Can you give some examples? I am working on PCG, and I'm very interested in creating an ML solution, but even the contemporary researchers seem to only be able to generate barely playable Super Mario levels. Every commercial title that I know of uses traditional techniques

[@MementoMori_2070]

⁠@@ansersoftware4463maybe The game Returnal idk

[@MementoMori_2070]

Nice pic. Run into any mind flayers lately 😅

[@ansersoftware4463]

@@MementoMori_2070 Unfortunately doesn't seem so, I found their GDC presentation about procedural generation in Returnal, and they are not using generative AI, just good old rules-based algorithms and RNG.

[@MementoMori_2070]

@@ansersoftware4463 what about no man's sky

[@DigitalXrisXros]

youtube algorithm : hi 👋 call me AL

[@dedster3164]

"This video is sponsored by Microsoft... " "Microsoft does not have any editorial influence on our videos" So, you saying what you got paid for an add and you can write whatever you want about this add?

[@KingTez1983]

That was short but straight to the point. Thank you Vox!

[@OwenAlekos-mh7yw]

Artificial interpreters, art officers internet. It's like the word human is humor in Spanish English. Artofficer being a real word describing how language and communication is art, and uses business office spaces and people representing business office space official guidelines. Thus, the day when people see the word internet being the same as interpreter and the same as intelligence, the conversation about what differences between language composition education profile preferences exists. Regarding how complicated language could be while seemingly being extremely direct and one way oriented. Thus, humor and metaphors being like an artists interpretation but also the business internet space. Thus, the concept of simplification of language is like a generalist approach to artificial intelligence and the internet. Potentially people are in conceptual artist interpreter programs for philosophy comparisons. Thus, pai, aip, API... API and geopolitical correctness of office spaces using valued formats of finance based communication; and accounting management. Regarding how philosophy and art can be interpreted through valuations of finance. Thus, generalist business finance for social philosophy could be API in the way that the internet interpretation and intelligence could be with a new sort of public domain. Like, API meritocracy credits for geopolitical philosophy and valuations; would be less of a language of interpretation or detailing or description or definition; but more like numerology and social office spaces. Regarding how the end of AI could be a metaphor for the ending of valuation and origins of humor or after philosophy or after artist documents. Like, having a public domain library office space for rehabilitation of communication capacity of retainable value. Similar to stating that the value of words can be bipolar or conflicting with itself like API without being online within the same grammar service program for typing or reading (in a lot of ways, as a game it makes communication correctly understood complicated)... Thus, discussing artificial intelligence a lot/ reading a lot about it is a confidence booster for interpretation of language; it's a big issue when people are often claiming, in society socially that the language used isn't proper enough for concerns to be resolved or finance accounting information to be correctly available. Like, it is a huge deal for making money and big money maker: the foundation of agreements.

[@benjamingeist1526]

Correction: you said “when that training data hasn’t been labeled and structured by humans, that method is called deep learning.” This is not true- it is called unsupervised (or self supervised) learning. Deep learning is the use of neural networks.

[@eloygonzalez840]

this channel is one the best in the whole youtube

[@kco3world]

The future's scary.

[@puffinjuice]

Who actually starts an email with "I hope this email finds you well"? What a pointless lead in sentence.

[@theavangelist]

Sponsored by copilot. The irony cannot be lost here Vox

[@HunterHogan]

Sponsors are inevitable, and I like that this video gave us multiple signals to distinguish the sponsor pitch from the video.

[@knoepfler34]

Yeah, they made it pretty obvious that they are sponsored. So I think that a lot of the criticism is quite unfair against Vox in that regard.

[@pedrosaraiva]

wait now we're calling out fill AI, too?

[@TheManinBlack9054]

always has been

[@martinkunev9911]

3:20 that's not what deep learning means

[@mwoods8988]

I think the reason survey responses leaned away from AI is because AI isn't that important for most of the technology listed, especially for the subset of features each respondent probably uses. Sure, if it's going to identify who's at your door, that's probably AI. But, if it's going to count how many steps you walked today, that doesn't need AI.

[@hoshismoodswing5309]

I'm not kidding. I'm a little bit scared 😢

[@YouTubecommentor284]

I can not remember the last video from vox that I found really interesting. The edit is there, but there is no heart anymore.. This used to be my favorite news channel.. Now it is just ads...

[@chrischy]

If you may stop calling bots and LLM as AI then I would be firm with it.

[@nithinravi9600]

“Microsoft’s safe AI principles” - sure

[@Noodles4L]

ooooooooooooof, I learned a lot with this one.

[@WarmWeatherGuy]

I have marked all "spam" as "not spam" in both my yahoo and gmail accounts, for many years. They keep marking some email as spam. It doesn't learn that I consider every email to be "not spam." This is not very intelligent.

[@lucasmonteiro1504]

The definition of deep learning is not quite it. There can be deep learning with labeled data, and there can be traditional machine learning with unlabeled data

[@Mrkiki0209]

The distinction between non generative machine learning models like what we use for an email spam folder and generative "AI" is really important here.

[@vasusharma8066]

Vox straight up getting the definition of deep learning wrong is extremely funny yet sad.

[@noizy514]

2023: marketing re-labels everything done by a computer as "AI" 2024: "look at all these great advances in AI. are you ready for the future?

[@Macsooder]

I find the idea of increasing AI use in our everyday lives as disturbing for several reasons. Not only does it play the risk of continuation and promotion of social inequalities, at some point with machine learning and self teaching, researchers can sort of stop understanding how the machine actually thinks. That can give us AIs where their exact thought processes are not as clearly known as they should be. I also wonder what the long term effects would be on future generations regarding critical thinking and the ability to be creative and solve problems. Don't get me wrong, AI has its uses and can be quite beneficial, but we need balance.

[@unbeatablesachin]

In all these were algorithm stands?

[@neociber24]

Algorithm are just the steps to do something, an AI itself is build with algorithms