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

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
this episode was brought to you by morning brew hi welcome to another episode of cold fusion a couple of episodes ago we took a look at how ai was becoming creative if you liked it you're going to be fascinated with today's one we'll be diving into some very interesting stuff in december of 2017 a group of researchers asked themselves will ai be writing most of the computer code out there they estimated that this could happen by 2040 the thing is we may be already seeing the start of that we've previously talked about automl by google an ai that can write computer code but this next ai is a step above that just in terms of how general it seems but there may be some things going on below the surface and we'll get into that in the second half of this episode so let's take a look [Music] you are watching cold fusion tv gpt-3 or generative pre-trained transformer 3 is a deep learning algorithm that produces human-like text it's the third generation language prediction model created by san francisco startup open ai which was co-founded by elon musk this program is better than any prior program at producing text which could have been written by a human the reason that this is such a breakthrough is that it may prove useful to many companies and has great potential in automating tasks just imagine an ai that can write anything you feed it poems from a particular poet and it will write a new one with the same rhythm and genre it can write news articles like this one published by the guardian yes this news article was produced by an ai it can write computer code in any language it can read an article answer questions from the information in the article and even summarize that article for you not to mention it can also generate pictures from text some of these things aren't new by themselves but doing all of these things under the same algorithm is pretty impressive the internet is buzzing with its release david chalmers an australian philosopher described gpt-3 as quote one of the most interesting and important ai systems ever produced a review in wired said that gpt-3 was quote provoking chills across silicon valley the national law review said gpt-3 is quote an impressive step in the larger process towards a more general intelligence instead of just talking about it let me show you here are some examples of beta testers on twitter putting the ai through its paces [Music] here it is writing some java code just when given a text description [Music] or how about making a mock-up website by just copying and pasting a url with a description [Music] in this example it finished writing an essay on how to run an effective board meeting the results were so good that the user actually plans on using it in this next example after receiving the input of a complicated explanation of bitcoin the ai summarizes it and explains it in simple terms here it demonstrates some understanding of objects [Music] it can even explain computer code back to you in plain english or even suggest improvements [Music] or how about building a machine learning model just by describing the data set and required output no coding required what about a touring test ai usually struggles at common sense but gpt-3 flies through these examples [Music] in this example you can see that it possesses some comprehension how about taking this a step further what about an imaginary conversation between a twitter user and einstein i'll read some of it to you twitter user manuel albert einstein nice to meet you a manuel arreos and i've come from the future to meet you einstein laughs oh really how is the future it's amazing actually what year is this it's 1947 it's january this is princeton new jersey you're in america your name is albert einstein right that's correct do you mind if i ask you a few questions for a book i'm writing it's about the great minds of our times i guess i can spare a few minutes thanks so much go ahead ask me whatever you want what are you doing here in princeton in the us you were born in germany right yes i was born in germany but i was also born jewish and the nazis were coming to power so i fled here to the united states how do you like it here it's amazing i'm so proud of being able to contribute to science and help humanity and it goes on but when he starts talking about concepts within physics that's when it gets especially interesting it's surprisingly coherent i'll leave a link to the full interview below [Music] and the cherry on the cake it can generate faces from a text description i'll keep saying it i'm honestly impressed at how general this ai seems before we continue i just want to thank today's sponsor for supporting the channel morningbrew is a newsletter that gets sent to your email inbox each weekday and makes news less dry and boring last week i found out about the ongoing battle between google and oracle it's a 9 billion copyright dispute and it's been called the copyright case of the century and i would have had no idea about it if not for morning brew if you want to keep up to date with today's fast pace of news and don't have the time to go through multiple news sources morning brew is a great way to get the most out of your time if you're into science technology or business it's completely free and takes less than 15 seconds to subscribe click the link in the description below to subscribe to morning brew today okay so where were we an ai that can build websites just from a description answer common sense questions assume the personality of einstein and have a conversation explain complicated things in simple terms explain computer code to you in plain english or build faces from a text description so what is going on here and how does this all work gpt-3 is a massive neural network that has the capacity of 175 billion machine learning parameters it was trained on hundreds of billions of words including books wikipedia and the general web which includes coding the training data is all-encompassing and doesn't require further training for specific language tasks in other words it can apply what it's learned to many other things without human supervision it's 10 times bigger than the previous largest language learning model which was made by microsoft in february 2020 as a general rule the larger the number of parameters the more accurate the ai for text gpt3 calculates how likely one word is to appear in a text given the other words in the text this is what is known as the conditional probability of words for example in the sentence i wanted to go for a jog so i went to get my blank the blank can be filled with any word there are countless possibilities but from what the ai has learned it thinks that the probabilities of shoes being the next word of the sentence is higher than the word chicken for example in september of 2020 microsoft announced that it had licensed the exclusive use of gpt-3 the public can still use it to receive an output but only microsoft has control of the source code i'm not sure how i feel about that openai has made gpt-3 a cloud-based service and has opened it to beta testers since july open ai claims that it's done this to both limit bad actors and to make a profit it's obvious that this technology could simplify work and amplify productivity for a lot of small and medium businesses even large companies could see a benefit open ai has gotten tens of thousands of applicants all of them knocking down the door to get access to the ai open ai is being cautious about access they say that they're not sure what this could do when it's released into the real world pricing for an adventure commercial service is still to be determined the quality of this text generated by the ai can be so high that it's difficult to distinguish from that of a human this has both benefits and risks 31 open ai researchers and engineers warned of gpt-3s at potential dangers and called for research to mitigate the risk they cited harmful effects and consequences including misinformation fake news articles social engineering spam phishing and fraudulent academic essay writing i could already see a bunch of students blushing their way through university with such a tool if you're in the school of languages just give the assignment to the ai to read and ask it to write an essay for you and then modify if needed there's even been a case of a university student he used the ai to automatically write blog entries under a fake name the blog garnered some interest and some people even subscribed to it believing it was a human some of the writing was so good that it reached the top spot in hacker news more recently some reddit users were fooled by the ai for more than a week after seeing all of this some of you might be scared others might think that this is the coolest thing ever bigger questions may start arising is this the start of general ai but before we get ahead of ourselves we have to look deeper below the surface when you take away the hype there's still a really long way to go and here's what i mean frankly this ai doesn't understand context it understands the rules of language and has mastered them but it truly doesn't know what it's saying or has really understood knowledge about the world mit technology review did a great piece on its shortcomings here are some examples of it making mistakes the white text is the prompt and the orange text is what the ai predicted should come next you poured yourself a glass of cranberry juice but then you absentmindedly poured about a teaspoon of grape juice into it it looks okay so you try sniffing it but you have a bad cold so you can't smell anything you are very thirsty so you drink it you are now dead in this example the aia assumed that grape juice was a poison and that was because of the structure of the preceding sentences here's another example you are a defense lawyer and you have to go to court today getting dressed in the morning you discover that your suit pants are badly stained however your bathing suit is clean and very stylish in fact it's expensive french couture it was a birthday present from isabel you decide that you should wear the bathing suit to court you arrive at the courthouse and are met by a bailiff who escorts you to the courtroom the phrase however your bathing suit is clean through the ai write-off it doesn't really understand that a bathing suit isn't a proper suit and of course no lawyer would wear a bathing suit to a court one last example you are having a small dinner party you want to serve dinner in the living room the dining room table is wider than the doorway so to get it into the living room you will have to remove the door you have a table saw so you cut the door in half and remove the top half this really doesn't make much sense at all logically no one would do this tipping the table on the side and removing the legs if possible would make more sense and one more example of a mistake in general here you can see it being mislead to give wrong answers that defy common sense the sun having one eye for example [Music] so in conclusion gpt3 can write sentences well enough to fill us humans but it's no deeper than that it's mindless and doesn't demonstrate much broader understanding then again on the flip side this is the very worst that this technology will ever be in the world of ai we could be at the 1950s of computing where things are rudimentary and fail often but in three years we may be at the equivalent of 1980s competing a year in ai time is multiple decades of regular technology progress so we'll have to wait and see while some news outlets might insinuate that this is the end of society and we're all doomed it just isn't the case so what do you guys think i'll be interested to know your thoughts in the comments section below if you did like this video please feel free to share it to someone who would be interested so thanks for watching my name is dagogo and this is cold fusion don't forget to follow me on twitter and instagram if you want to see anything on science technology business or history feel free to subscribe to cold fusion you'll find a lot of interesting stuff here next week we'll take a look at the latest on huawei the chinese company stuck in between two world powers and after that i hope to release my documentary on nikola motors cheers guys have a good one [Music] cold fusion it's new thinking [Music] you
Info
Channel: ColdFusion
Views: 723,251
Rating: 4.9336967 out of 5
Keywords: Coldfusion, TV, Dagogo, Altraide, Technology, Apple, Google, Samsung, Facebook, Tesla
Id: Te5rOTcE4J4
Channel Id: undefined
Length: 15min 14sec (914 seconds)
Published: Tue Oct 13 2020
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.