From Essays to Coding, This New A.I. Can Write Anything

13 okt 2020
692 711 visningar

Sign up to Morning Brew for free today here:
--- About ColdFusion ---
ColdFusion is an Australian based online media company independently run by Dagogo Altraide since 2009. Topics cover anything in science, technology, history and business in a calm and relaxed environment. GPT-3 certainly is something.
ColdFusion Merch:
If you enjoy my content, please consider subscribing!
I'm also on Patreon:
Bitcoin address: 13SjyCXPB9o3iN4LitYQ2wYKeqYTShPub8
--- New Thinking Book written by Dagogo Altraide ---
This book was rated the 9th best technology history book by book authority.
In the book you’ll learn the stories of those who invented the things we use everyday and how it all fits together to form our modern world.
Get the book on Amazon:
Get the book on Google Play:
--- ColdFusion Social Media ---
» Twitter | @ColdFusion_TV
» Instagram | coldfusiontv
» Facebook |
Einstein interview:

Hiatus - Cloud City
Winter Flags -Winter Flags
Burn Water - Nostalgia Dreams
Sworn - Sorry
Croquet Club - Only You Can Tell
Vesky - Remembrance
Brock Hewitt - Peaceful
Cahb - Secrets
Burn Water - Dissapear
» Music I produce | or
» Collection of music used in videos:
Producer: Dagogo Altraide
Why are you still reading?

  • Still waiting on the first person to catch the audio Easter egg. Also a correction at 2:48, the code is JavaScript not Java. Enjoy! PS: If you’re interested in automation and it’s societal effects, I have a video on UBI here:

    ColdFusionColdFusion6 månader sedan
    • @The Fixer Now tell me what it is about modern neural networks specifically that makes it impossible for AI to be "aware".

      fiso64fiso64Månad sedan
    • Showing a few cherry-picked examples of GPT-3 making mistakes doesn't mean it's "completely clueless". We don't know what consciousness is, we don't know how far away it is from being conscious, and we don't know whether or not it understands anything. This is not a critique for your video specifically. A lot of people are downplaying GPT-3 without bringing any good arguments.

      fiso64fiso64Månad sedan
    • @Joe Bentlee awesome! Took about 15 minutes but it reallyworked!!

      Joziah RemingtonJoziah RemingtonMånad sedan
    • Dont know if anyone cares at all but last night I hacked my girlfriends Facebook password using SicZine. Find it on google if you wanna try it

      Joe BentleeJoe BentleeMånad sedan
    • @Frances Nongkhlaw To know about a soul is one thing and to comprehend as to being of that essence is another.

      Isaac PerezIsaac Perez2 månader sedan
  • The "what is a factorial" at 4:20 is comical! What a trivial reply GPT3 gives.

    George BrantleyGeorge Brantley7 timmar sedan
  • Something you forgot to mention: it is computationally impossible for a program to write a non-trivial program. Don't remember the exact mathematical theory that proved this but do remember that theory was deduced from Cantor's Incompleteness Theorems and the Halting problem. However, we can approximate which is what current research, including GPT-3, focuses on. So in essence we'll never truly automate programming.

    macroxelamacroxela19 timmar sedan
  • I see that it should be taught manners and occasions ! And the correlation between the m!

    Deathto SpiDeathto Spi4 dagar sedan
  • Where can I read the rest of the Einstein interview??

    Graham HartGraham Hart7 dagar sedan
  • Thanks I Hate it! :D

    Purity is a state of mindPurity is a state of mind11 dagar sedan
  • If they release GPT3, hacking will be far more easier

    Antonio FullerAntonio Fuller11 dagar sedan
  • Looks weak, I've seen better

    Antonio FullerAntonio Fuller11 dagar sedan
  • Superb! Yep, A.I. can never have a spirit, soul or consciousness. Ankh you so much.

    KingAbdullahKeyKingAbdullahKey11 dagar sedan
  • pft. i think AI is going to be making your youtube videos before it does my programming! :D

    GabrielGabriel12 dagar sedan
  • Can't wait to see it being released to the public and go down the same path as Microsoft's Tay, but even faster. :-)

    LaughingStock71LaughingStock7112 dagar sedan
  • Best contend on SEworld. Have seen all of your videos here. Thanks so much for sharing this with us

    SennyahSennyah13 dagar sedan
  • Ok

    Karl Quinte SwiftKarl Quinte Swift13 dagar sedan
  • Well this reduces my envy of my software developer friends a tiny bit lol, as a civil engineer in a public utility no less....I’m valued with age and career longevity aaaand project management of physical ppl using software seems to be the extent of what will happen in my lifetime.

    M0 HereM0 Here14 dagar sedan
  • 2:47 - Fun fact, it's not Java, It's JavaScript, specifically React, but no worries, a channel like this one taking care of technical side of things doesn't have to know the difference...

    ShapelessShapeless15 dagar sedan
  • This is a very interesting topic indeed. Let us assume a computer that generates randomly ASCII characters. Applying temporal logic, the program will always ends up by producing the best comp[uter program or the best novel. Now to do that we need time, a lot of it. The question to answer is how do we reduce that time? We can reduce that time by increasing the CPU computational power and using what is commonly denominated as artificial intelligence to train the application by reducing the number of highly potentially failed attempt. There is no intelligence, only mathematical regression curve that compute the best path or interpolate using minimum distance. The program is not aware of anything, therefore there is no intelligence. Once the program will be aware of its existence, we will not name it artificial intelligence but an enemy that must be destroyed.

    ab bbbab bbb15 dagar sedan
  • 2075- My bitcoin has touched $10 million. Wow! AI to nearest robot- Just kick him hard and put him to work. This bitcoin is distracting him too much. Empty his account and put some dollars instead. Robot- unable to find suitable work for him. AI- OK, use him as a battery cell.

    R RR R15 dagar sedan
  • are you gp3

    S HoganS Hogan15 dagar sedan
  • Never writing another school essay again just gtp3

    Johnson JameJohnson Jame15 dagar sedan
  • OK so this seems to be just hype and nonsense. GPT seems to be just a kind of advanced auto-complete-it doesn't write stories it simply offers a choice of continuing words (or just word) based on what went previously. Most of the work still has to be done by a human, and it is slow-I can write articles faster.

    Archvaldor's Warcraft HacksArchvaldor's Warcraft Hacks15 dagar sedan
  • I am really really impressed!!

    I. DiarraI. Diarra16 dagar sedan
  • It’s smarter than median American for sure!

    Arthur MorganArthur Morgan22 dagar sedan
  • 4:35 Hah, frogs only have 2 legs dumb GPT-3. Also, dogs pant and lay down when they are hot. They don't get upset about the heat. Although hot is a synonym for angry, if someone said, "That dog is hot," then I think people would be thinking a lot of other things besides the dog being in a bad mood. Lastly, animals with 3 legs don't fall over, they get used to it.

    TT24 dagar sedan
  • I have been saying for along time the pelosi biden harris chuck McConnell obama the entire team are using AI with production so fast of long bills being worded tricky to americans and goving them to republicans on day of court . Not shareing and negotiations w both parties.she couldnt do all these bills herself that fast. I heard the computer told her to kill herself lol shes really evil

    Kimberly HKimberly H27 dagar sedan
  • Magic

    SSSUESSSUE27 dagar sedan
  • AI only make the rich getting rich, and the poor getting poor temporary, maybe a few decades, after reach the climax, AI will start indirectly reduce humans population, as humans has become redundant, and due to less job, humans will have less baby, less baby means less demand, less demand mean less production, in the end, it just a doom for humans, it is the best solution that make the world fair

    Teoh WSTeoh WS29 dagar sedan
  • Waaaaaait. Wasn't Elon Musk staunch anti AI in the AI debate between Zuck and Elon on the future potential of AI. And you say he's responsible for this AI now being bought by Microsoft. hmmmmm

    The Jobless CoderThe Jobless Coder29 dagar sedan
  • thats not "understanding" it's algo cross referencing search engines. nothing more than source gathering

    Charles SkotekCharles SkotekMånad sedan
  • The risk with AI is, that we will not make it to a strong AI (to destroy us), because AI tools like GPT3 and its successors will be abused as tools by people to bring humanity to the edge of the cliff.

    Jan RadtkeJan RadtkeMånad sedan
  • Showing a few cherry-picked examples of GPT-3 making mistakes doesn't mean it's "completely clueless". We don't know what consciousness is, we don't know how far away it is from being conscious, and we don't know whether or not it understands anything.

    fiso64fiso64Månad sedan
  • Don't worry, there is no career path that won't be affected by the AI.

    PiterixosPiterixosMånad sedan
  • As a procrastinater I'm dying to get access. I have so many online and offline business ideas but my glioma limits my ability to finish tasks or goals. I would choose this over litteraly anything. Shame its not open source. I can delegate the things I struggle with offline so using gpt3 online would be a dream come true.

    B WolfB WolfMånad sedan
  • Of course we are all doomed. That is how life works. We are born, we live a while, and then we die. C'est la vie!

    sam1953sam1953Månad sedan
  • I'm excited about this sort of thing. Even though it means students can cheat exams, it's an example of how the educational system needs to be changed in order to facilitate current technologies. For example, ever since Google and Wikipedia became so prominent in us researching various topics, it's no longer about memorizing stuff. It's about knowing *what* information to use. In other words, it's about *how* you think. Computers and eventually AIs will be able to come up with extremely amazing stuff for us to watch and/or read about. However, at least for the time being, it's up to us to set the parameters. Until that happens, human intelligence will still be tantamount.

    Karl Marx stole my bikeKarl Marx stole my bikeMånad sedan
  • say good by to jobs by 2045 if not 2050 self driving and AI/Boston dynamics we are heading to a world where you are paid for being a human and if you are a robot you just do the work we will have to be care full to not make them sentient that will be only thing we have to worry about that and keep learning how to remake it if it dies or gets hit with an EMP all in all we will have to change the way we see people as low human sub human high human its like that now based on your smarts and money and political influence it will still be a thing just a pebble instead of a mountain

    wolf rain88wolf rain88Månad sedan
  • So coding makes coding obsolete.

    obimk1obimk1Månad sedan
  • AI might be able to govern a country in the near future. I mean if it can win at chess the best human being at it than it definitely can do a better job than the people in the parliament.

    Sergiu BaracSergiu BaracMånad sedan
  • Why are you still reading? 🤣🤣

    Muhammad Zubair ALiMuhammad Zubair ALiMånad sedan
  • See those square shoes at 11:06?

    keithy077keithy077Månad sedan
  • Too hyperbolic. The Turing test was shitty at best. And so goes for the use of terms like "it understands" (it doesn't, that's an anthropocentric way of describing a mere result of an input, in the same way that your PC doesn't understand the use you make of it). Etc. The fact that it can cheat lazy readers onto believing those articles (etc.) were done by humans (and the fact that of course and lately, Internet writers are not especially clever and get worse year after year) doesn't mean a such a big deal. OF COURSE that it can facilitate some task, but the A.I. acronym is still and euphemism rather than a reality.

    magnvssmagnvssMånad sedan
    • Someone didn't watch the full video :)

      ColdFusionColdFusionMånad sedan
  • AI will generally make idiots capable of simple things, and competent people capable of more advanced things.

    KyrieruKyrieruMånad sedan
  • I, for one, would like to welcome our A.I. overlords

    charles farrarcharles farrarMånad sedan
  • Finally, we can create new Emo songs like they were written in the 2000’s. #EmosNotDead

    ii-ViCioUs-uK__ii-ViCioUs-uK__Månad sedan
  • *This is Black Magic*

    RyuzakiRyuzakiMånad sedan
  • *But you need someone to keep a problem which AI needs to solve* I rather be that person

    RyuzakiRyuzakiMånad sedan
  • Literally anything Elon touches is a breakthrough...

    Mark JohnstonMark JohnstonMånad sedan
  • WTF!? Microsoft has control of the course code? They're scum of the earth. DO NOT TRUST THEM.

    UnbelievableUnbelievableMånad sedan
  • At 4:38 I was playing a program written in BASIC that could emulate speaking to another person in 1979 - any of the AI examples so far could have been written in BASIC or C or Python and would have been simple to do. There's nothing much good here about AI, it's all ANCIENT stuff that would only impress non-programming people (lots of them around).

    Gort NewtonGort NewtonMånad sedan
  • Thanks for pointing out that tipping the table on its side would make more sense. I guess I’ll put my table saw away.

    Paul SPaul SMånad sedan
  • find that, just as a large transformer model trained on language can generate coherent text, the same exact model trained on pixel sequences can generate coherent image completions and samples.

    Honvai HonvaiHonvai HonvaiMånad sedan
    • Add Camera Now we have Bender.

      Honvai HonvaiHonvai HonvaiMånad sedan
  • Plot twist: this youtube channel is run by A.I.

    RoyRoyMånad sedan
  • Read that A.I. article it is, terrifing.

    40kBrit40kBritMånad sedan
  • "Here it is writing some Java code" *Shows it outputing React Code, which is Javascript based"

    TheBrazilRulesTheBrazilRules2 månader sedan
  • Why would you stroke a stuffed animal?!!!!!!

    TheBrazilRulesTheBrazilRules2 månader sedan
  • The nutty driving critically enter because swimming symptomatically crush barring a distinct witness. breezy, ancient swallow

    pudpud kirkpatrickpudpud kirkpatrick2 månader sedan
  • you all have no idea. it IS already here since a long time. smarter than you think and controlling us. gpt-3 is just a small peanut fraction in its cake.

    ASHiGA DesignsASHiGA Designs2 månader sedan
  • The mistakes are just as frightening as the successes lol

    Jeffery SternerJeffery Sterner2 månader sedan
  • "The sun has one eye" *Iluminati backsong has been played*

    Ariq HaryoAriq Haryo2 månader sedan
  • Usually I don't comment but bruh your music is DOPE! 🔥

    Wealth DecodedWealth Decoded2 månader sedan
  • The day AI can tell who farted, that’s when we gotta start worrying y’all.

    RUFU5RUFU52 månader sedan
    • by far the best comment here.

      nickrd182nickrd182Månad sedan
  • For me it sounds like GPT-3 is a child still learning about the way things work.

    ViSiD _ViSiD _2 månader sedan
  • This is scary shit

    jay methodusjay methodus2 månader sedan
  • Does not impress me to very much but we are still doomed because we actually have faith in this false idol way more than we will the actual God! Sinners burn in hell right? Forgive me I was forced to attend catholic school for my entire schooling! lol... I don't believe it & neither do I believe technology can save us from ourselves! That if fucking ridiculous! Keep cheating life & you will find out the consequences the hard way!

    Scott IsaacsonScott Isaacson2 månader sedan
  • Why the twitter link for einstein interview not working?

    Vishruth RaoVishruth Rao2 månader sedan
  • Oh now i see. Guardian articles are written by AI thats why they are so extremist and biased.

    Marco MöngkeMarco Möngke2 månader sedan
  • Fictional/artistic writers will remain because art is about individuals expressing themselves and people are generally interested in people. It's not just about the final word on the page. In general AI will find it's place because people won't make themselves completely redundant. Important is always the technology people want and not the one that is theoretically possible.

    James CooperJames Cooper2 månader sedan
    • I've seen ai produce simple background music and paint, out side of that it still struggles with storys, so my future profession is safe, for now until it gets sooo good that it can outplay us all at the game of writing/storytelling, utilising each and every style possible to suit the consumers taste. Imagine something so perfect ever evolving and evermore capable able to deeply understand the great masters of the craft in a much shorter time than us, pumping out works endlessly never tired and the quality stays excellent all the while. It would be a metaphorical bloodbath. A good clump of people tend not to care as long as it's good or scratches the itch they have. Plus if im hung up on a certain artist or style the ai will be able to nail that too on command. Heck some might go as far as analysing one authors works and their personality and have the ai live as that keeping the work going beyond their death They already do holograms of dead celebrities at concerts.

      Ahdh WjdueAhdh Wjdue14 dagar sedan
  • Will GPT-4 be the perfect spy? All of these possibilities to understand test and thus language and to bring them into a meaningful context are frightening when you think about the scalability

    AkumaAkuma2 månader sedan
  • Very interesting. However, I have developed a method of repairing human tissue in the ambulatory patient. (pancreas, kidney, brain). I have increased the IQ of "retarded" patients...and even repaired my own brain tissue loss I sustained after several strokes. Using the right agents humans organs can be repaired (without transplant surgery) and IQ gradually increased. This technique can possibly give robot makers a run for their money. Don't count humans out prematurely. Some of this information is in my last textbook ("Disorders of Blood Sugar 2013" Also enhancing human function and intelligence is a lot cheaper and building massive machines. And...fixing kidneys is cheaper than dialysis.

    Roberto IllaRoberto Illa2 månader sedan
  • This is definitely a company you should pay attention to when the ipo drops it could make you a fortune

    Dougie JonesDougie Jones2 månader sedan
  • I think that... in the end it will be impossible to make AI self-aware. It will get better and better at fooling us into thinking it's self-aware. But in the end it will be impossible. So unless we program it to be a retarded genocide maniac, AI is not a threat to us. Our errors while making it are. I just hope that one day we will have our own AI robot each, and it will work for us so that we can just fck around like drunken apes we are. We want to be redundant, to be taken care of, to be free of labour. So that we can all pursue our own passions and hobbies and live a happy life without it's burden. To reach for the stars, to make and create without pressure only with joy of the act itself.

    GawędziarzGawędziarz2 månader sedan
  • The "cut the door in half" bit is a great example of how dangerous general AI can be if not set up properly. If given a problem, it might do all sorts of unpredictable things to solve it, like killing all humans for resources, for example.

    Rafael LoboRafael Lobo2 månader sedan
  • Louie Louie, oh no, you take me where ya gotta go ... yeah, yeah, yeah, yeah, yeah, baby ... Louie Louie, oh baby, take me where ya gotta go ... #AiRobot ...

    Sparky O'brienSparky O'brien2 månader sedan
  • hopefully my toaster doesnt start cutting my dooros in half.

    Dante DelodenDante Deloden2 månader sedan
  • It is very cool to learn tho. But it is pretty scary if you think that in the future an AI will program other AI’s

    P1N3C0N3 DДN0OBP1N3C0N3 DДN0OB2 månader sedan
  • There is already social engineering in a controlled society. U can not place blaim on GPT-3 FOR ITS POTENTIAL on social engineering. We also feel A.I. is a threat to humanity but humans have been a threat to humanity for many years and getting worst with time. Hopefully i won't be around for the technocratic society that so many anticipate. The future will be very ugly for the underpriviledged but in the long run those on top of the pyramid structure will also perish. Where there is hate and destruction extinction is sure to follow.

    Milton RiveraMilton Rivera2 månader sedan
  • What I am curious about is if an AI is so smart to write such nice article. How come, translations and grammar correcting softwares are still so so so trash that their translations are sometimes barely understandable and the grammar correcting software sometime fail to correct many grammars.. 😳

    Dorky Cuty PiDorky Cuty Pi2 månader sedan
  • AI programmed to understand pleasant, safe, sustainable urban environments, connected to a 3D printer. b00m, American suburban sprawl hell solved.

    Patrick CarnahanPatrick Carnahan2 månader sedan
  • This is a bluff by Coldfusion. A i as shown at the beginning is far more capable than the ending. But like all great things it definitely is more terrifying than Nuclear Warfare. Elon Musk : Jan 2021

    AndrewAndrew2 månader sedan
  • Not sure if we are all doomed but we will eventually all die. We have advanced and progressed enough to know that the earth, the sun, the solar system and the universe will eventually fade away some day. And even though we have a pretty good idea on how far away these events are, they are so far into the future that they will not really impact the continued evolution of life on earth for our lifetimes and for the lifetimes of our descendants for thousands of generations. Universes may have continuous birth, growth and death cycles. The universe may be a cycle that can regenerate itself and eventually regenerate life. Ever since life first developed and evolved on earth, all life has had to compete for survival. Ever since humans started to dominate life on earth, they too had to compete for survival. There has also been a tendency for humans to compete in trying to conquer, dominate and control other humans and get them to be servants and slaves. Using force including beating, torture and killing to compete is also something that humans have had a tendency to do. To help them achieve these goals, humans tend to organize themselves into groups such as tribes, communities, societies, kingdoms, empires, dynasties and nations along with organizations and associations with a hierarchy and ranking level structure of command such as corporations and political parties. Humans have also created caste and class systems and have an amazing talent to tell fascinating and conflicting creation stories that have also been used to conquer, dominate and control other humans sometimes for the better but often times for the worse and to justify bad behaviours such as slavery. For most humans, life has always been a struggle and a competition for survival. If there is any chance for humans to live on earth in fair and peaceful ways without abusing, killing or dominating each other and destroying earth’s environment in the process, humans will have to learn that: conquering, killing, oppressing and enslaving other humans along with, stealing, cheating, lying, spying, deception, greed, betrayal including violence, coercion, extortion, corruption are never beneficial in the long run. Humans also will need to learn and understand that reason, rationality, logic, honesty, trust and responsibility along with the understanding that the well being of others benefits everyone and is the basis for morality are the best ways we know of so far to live in peace and fairness. The idea of democracy with systems of law, order and justice also need to be encouraged and promoted and when trying to learn and understand the nature of things and solve problems the scientific method needs to be embraced. I am hoping that the explosion of information technologies like the internet and artificial intelligence can help and inspire all human life on earth to live together in fair, peaceful and sustainable ways. Unfortunately information technologies like the internet and artificial intelligence can also make it easier to spread false and fake information and help inspire division and hatred against other humans, other cultures, or other nations. The outcomes are really never predictable.

    DannyDanny2 månader sedan
  • what app did he use at 3:12?

    A FA F2 månader sedan
  • Always thinking for a while now,if every single job are replaced by the AI who would buy the product when everyone don't have a job to get money to buy the products

    Glend Junus Lodo LiGlend Junus Lodo Li2 månader sedan
  • Create some algorithm parameters for economy and government. Oh and energy, new battery technology, solar, wind, they are probably already doing this, add all of Nikola Tesla’s experiments to a parameter

    cory ccory c2 månader sedan
  • My thought: start your channel on LBRY, please

    Shahram JavidniaShahram Javidnia2 månader sedan
  • The code was Javascript aka ECMA Script not Java.

    Jens BröcherJens Bröcher2 månader sedan
  • GPT3 hurt my feelings.. i wanted to date her, she seemd perfect for me but she kept making this story of how i should be more patient till she gets ready for dating..

    Albert PecolajAlbert Pecolaj2 månader sedan
  • Gasi internet , motaj kablove !

    mihael čituljskimihael čituljski2 månader sedan
  • We humans are doomed, AI is going to win. Regards AI👁👀

    w0bblyd0inkb0inkw0bblyd0inkb0ink2 månader sedan
  • looking forward to see AI leave comments on youtube...

    Amit Singh PariharAmit Singh Parihar2 månader sedan
  • That last shortcoming example of "removing the door" is the exact reason why everyone should be cautios of AI. Imagine an AI that could influence the physical world making such mistakes on real tasks

    Osa AimolaOsa Aimola2 månader sedan
  • Man may be plunging headlong into obsolescence, it seems, very much like the impact of machines on labour in the industrial revolution if indeed that revolution ever paused. Nice to know the tower of Babel will be waiting at destination with translation machines.

  • i love how they’re like “we don’t want bad actors to get ahold of it” *hands rights exclusively over to microsoft*

    sonic the hedgegodsonic the hedgegod2 månader sedan
  • It will not really need to understand ... only get better and better at appearing it does. When it can fool us every time how will we ever know it's really just a dumb machine?

    Martyn HaggertyMartyn Haggerty2 månader sedan
  • Did no one watch terminator. This is two dangerous.

    Jeremy DeanJeremy Dean2 månader sedan
  • Coding no , because u need programmers to code the ai and clients to give clear requirements hahah.

    Sommyarup RoySommyarup Roy2 månader sedan
  • Yes it can write its own version of the human genome. Which i remind you in its essence is computer code. Every atom in its binary position.

    Edward WoodsEdward Woods2 månader sedan
  • But can it write "Hello World"?

    Dinesh waktiDinesh wakti2 månader sedan
  • My intelligence is already artificial..

    Jemel BennettJemel Bennett2 månader sedan
  • Looks like weaksauce👌

    MCES LEXMCES LEX2 månader sedan
  • thinking 2021 is a bad time to learn how to code

    Ak DarkLightAk DarkLight2 månader sedan
  • 11:57 sound allot like AI Dungeon to me

    Greg Del PilarGreg Del Pilar2 månader sedan