ChatbotGPT is a new artificial intelligence programme designed to simulate human conversation and tackle complex questions.
(Subscribe: )
It's made by Open AI foundation, a tech-startup co-founded by Elon Musk, and it draws on text taken from a variety of sources on the internet and its creators say it has learned how to answer academic questions, and even sometimes admits when it's wrong.
We've done an interview by putting questions to the chatbot, and then generating a voice for it using different software.
We asked the Chatbot GPT whether fears about A.I. threatening the human race are well-founded.
———————–
Follow us on Instagram –
These questions were pretty basic. It can give pretty good answers to complicated technical questions, as well as generate all kinds of content. Its knowledge base is huge, almost any fiction or non-fiction topic you could think of. You can ask it how two unrelated things or concepts relate or compare against each other, ie unlikely to be copied verbatim from some existing web page. You can tell it things like “you are now a text video game where you give me options” with some extra details and it generates a custom text adventure for you on the fly. You can dump your own writing or documentation into it and then make a request, and it will seemingly understand it all and give you a helpful answer.
But no, we get to hear it talk about itself and what it thinks about gender
“But no, we get to heart it talk about itself and what it thinks about gender”.
Welcome to Channel 4 television channel!
@young hokage You say it is smarter than “you and I” but can it solve a fifth order differential equation with constant coefficients?
I am guessing not. Now I can. So is it really smarter than you and I? Smarter than you may be, but not I.
@Dean G 😂😂😂😂😅
🙇🏿🙇🏿🙇🏿🙇🏿😅
Says that a woman is a person with female reproductive organs etc and then when asked is a Trans woman is a woman says yes to that and being a woman is not solely determined by gender or anatomy …EPIC FAIL!! Completely contradicted itself🤣🤣🤣🤣🤣🤣🤣🤣. Just shows when the data set is garbage then garbage is all that comes out!!
My take away on this is the same as it has been. I am a software developer and I know that there is no 100% perfect software because humans are not perfect and they create the software. An Ai really isn’t any different than any other software … just it’s interface has changed to be more personable. It can only base it’s response on data that it has available to it, which again is governed and provided by humans, so it is not perfect and will remain bias for that reason. This is how it has been since the invent of computers. You get dozens of software updates regularly to fix something. Why? Because it’s not perfect. And it will never be perfect as long as it’s created by humans.
That said, are there places that would be well suited for Ai? Absolutely. But there are also things that should not involve Ai. Even if you develop an Ai capable of emotions, those emotions will only mimic human emotions that it learns from. Every human has different emotions and perspectives based on their life experiences which literally means that no one person will actually “feel” the exact 100% same way as another human regarding any specific topic. It’s not possible. A simple difference in blinking of one’s eye at a different time can be enough to create a difference. Everyone’s brain processes information at least a bit differently. For this reason, it will be literally impossible for an Ai to fully understand every person’s human emotion… and in some situations this will be a very important factor.
An Ai can be trained to recognize human body language… and an entire assortment of things humans do automatically without even putting effort into doing. However, every interpretation of what is being observed will be based on the data that it has. That is a huge factor in my opinion. Because there are some things we just know. We can’t always say how we know, we just do. I am a very logic based person… but I also rely on instincts and gut feelings in certain situations. I can’t say something is wrong with someone else based on actual data always, I just know that there is something wrong. Ai doesn’t have that sense and I’m not actually sure that it can.
So, while I’m excited about the advancement of technology, I’m a geek, I’m not so excited about how people are using it or want to use it to replace some things that are inherent to humans.
Productivity and social media use will be wild with ChatGPT
I guess……
I’m looking forward to a video when you explain how to use this AI to be more productive ! 😅
This is dangerous. This is the start of the demise of the human race. We must stop advancing and obsessing over technology
@Dominik Klon Easy. Just tell ChatGPT to write you a (#?) page essay on (insert topic here).
Full essay written in 3 seconds.
That is productivity.
If they could smooth out the voice and make it more natural, it I think we would be seeing something like what we saw in the film Her.
This interview just scratches the surface of the utility of the chatGPT model. I have found that by asking a series of questions and building up a contextual big picture as you go in a given chat thread about one topic of interest is likely to be rewarded by illuminating results. This is truly a breakthrough proof of concept AI model with myriad applications
What this suggests is that asking questions has greater value than the answers themselves.
@NewTube Channel It’s like a library, people who can’t read will enter it look around and leave. A person with some reading skills will flick through some books and leave. A really knowledgeable person could spend hours there perusing different books and leave with more insight on certain topics than he had when he entered.
@redshift Cool, what an irrelevant analogy. As a user/student there is no shortage of the ways to research material on a topic. This may or may not be a way (depends). It calls into question what is considered as the “gospel truth”. One of the major issues here with the bot is that you have to limit your bias in a context that’s full of implicit biases. It means the associations between results are highly subjective. The bot is unable to make any arguments, yet is designed to go out of its way to pretend as if it can. As a rule of thumb, as a user before jumping on bandwagons take a moment to fully exhaust your problem area. BS responses are worse than no responses.
@NewTube Channel I haven’t got it yet but I’ll never ask it whether I should take up keto diet or intermittent fasting.
However there are thousand of programming languages, If I need to learn a new one I’ll definitely ask it to write pieces of code for me so that I can analyse them. A smart person will intuitively know when to depend on it, and when to base certain decisions solely on the workings of his brain cells.
@NewTube Channel it can work with you on things. You can act as editor or executive and it really works.
Not too bad as an introduction to ChatGPT for a wider audience. Using a text to voice converter probably gave the casual viewer a slightly skewed view as reading the output is a more neutral way to appraise it. For anyone new to this tech, with any question if you hit the regenerate button it will come back with another answer, that over time will repeat key sections of previous answers. (This can be changed by amending certain settings on the web version.) So ‘how would you improve the world?’ etc. can be rerun until you are happy with the result. I’m not saying they did this for the interview, but they didn’t say this was one contiguous conversation.
It would have been nice if a few ‘creative’ examples were shown. I got a fairly amusing West Wing scene written in a TV script format by simply typing “produce a script for the West Wing featuring all the main characters set in the press room involving the Thanksgiving turkey escaping. Make it humorous and have a surprise ending“ It had CJ panicking that the press corps would find out and ended with President Bartlett demanding DEFCON 1. Although I don’t think Aaron Sorkin is out of a job yet.
Then I got it to write an AA Milne poem with Pooh Bear playing poohsticks – it knew how to play the game. A Larry David commercial for his own brand of toothpaste included his ‘Pretaay, Pretaay, Good’ catchphrase, but it was probably not too hard to add that. Six new plots for Fawlty Towers, with each episode featuring a famous guest star and a surprise ending, included a new chief to help Manuel was really Gordon Ramsey in disguise and another starred Hugh Laurie tuning out to be an undercover MI5 agent and not a petty criminal. It knew the basic premise of a 47 year old British sitcom, presumably from hoovering up the scripts in its training data. It also gave me very useful advice on how to write a good science joke and spat out ten of them, all a bit lame, but a few were funny.
‘It’ was rather good at writing an article stating that ER does equal EPR, citing professor Susskin, Einstein and others. Although it can’t have known that recently a wormhole was produced in a quantum computer at Google because it’s training data ends in August 2020. But as an article on the subject it was very impressive and completely correct technically, as far as I could tell as a layperson.
10 uses for a paperclip, a paper cup, and then a CD, first blank and then damaged, produced a set of sensible answers for storage and then some creative answers, including using it as a frisbee or a drinks coaster. (And that’s when I think you go wtf!, but it’s still a trick, it didn’t think the ideas up, it simply analysed patterns in other text – unless it was trained on how to repurpose damaged CD’s.)
It correctly told me how to subnet my home network, it wrote some working python code and it gave me a very impressive answer when I asked “explain the meaning of 2001 space Odyssey in five levels from a child’s understanding to a university professor“. It was really helpful when I told it my dishwasher was leaking water and it told me how to fix it, while advising I should seek help if the advice didn’t work.
As I’m sure many have done, I took large chunks and small chunks of the text and Googled the phrase output and not once was anything returned. So every phrase is unique and this must pose a problem in the future for any article or essay written as part of a school project for example, as I don’t think you could tell it was from an AI. Recently, tests have been conducted with experts and other AI’s to try and determine human or AI sections of a submitted article. Neither of them could be certain of the original source.
Some of the above text is from ChatGPT’s output …
This really shows alot about our society on the proceeding questions on 8:20
Fascinating! I told my girlfriend about it and she just goes ‘oh yeah and carried on nagging me about washing up or something!
@Mr Tambourine Man same 😂
@Gianpa haha 😄
@Mr Tambourine Man lol 😆
I’m literally using this everyday. If you are a software engineer it’s truly a game changer. It pumps out code for you like clockwork.
@JYG In nearly 60 years of coding I’ve been in the habit of “reviewing” my code by running it on test cases, trying to include as many corner cases as I can. And I’m someone who has published extensively in the area of program verification since the 1970’s, so it’s not that I don’t know how to prove my programs correct, I’ve simply found testing turns up bugs at a greater rate per line of code than proving can. Simply testing while looking for corner cases is how codebots will outdo humans.
Shhhhhh!
@O L maybe your boss will lose his job too lol
@CZ who knows, you might be right
Heading down to singularity.
For me, this is awesome. I don’t care what flaws are visible to us. I admire the fact that we as a specie could come up with such gigantic innovation.
CBDC is the biggest tyranny the entire world has faced in a long time. We need to stand up. If not your freedom is so fucked you won’t know what hit you.
Seems to me that we have all overlooked the fact that Artificial Intelligence arrives from machine learning, and not from a few million years of evolution. The algorithm clearly stated that it cannot express emotion. Yet the answer on empathy suggested specific courses of action. Which if followed through would undoubtedly make the word a kinder, safer and and more productive place. We ought to remember something of our own physiology and never forget that our words and deeds often are driven by the vagaries of the amygdala, and not as we would imagine but the prefrontal cortex. Which in a world of WMD’s, pollution, and climate change, matters.
It is the handy work of God Almighty , From an Islamic perspective when HE created humans the angels questioned its purposes and God showed them the power of the mind of Adam the new creation by asking Adam to name every Angel names, animals etc….. and he was shown all the creations from start to end, it blew the mind of the angels who never knew what human meant. hence why when God ordered them to prostrate to Adam as a sign of respect, they prostrated to Adam but Satan the evil refused because he was arrogant. Other religeous books also speak of similar encounters.
But we human beings came from nothing?
@Islam Khan no, we came from dust.
I asked the GPT to come up with rules for components for board and card games I already have, and they were not bad at all honestly. Definitely produced some misunderstandings and design contradictions, but the average board game manual is Okay-to-Good, so it’s definitely impressive.
I would consider misunderstandings and design contradictions bad, but you’re the specialist.
the more specific you are in your prompts, the better your results
It’s simply a programmed response database based on input variables, not sure why everyone is excited or worried. It works because of smart people entering the data and is only as smart as those people are collectively. When it starts solving problems on it’s own that have not yet been solved by humans is when you should worry or get excited.
“I’m sorry if I’m sometimes incorrect in what I say, I can only work with the information I’m aware of” – if more people were like this the world would be a better place
@tutacat tinfoil is on discount at Walmart this week I hear.
😄😄😄🤣🤣🤣🤣
Yup, life would be fascinating
I am sorry Dave. I can not open the pod bay doors.
@EL dEL Do you do much reading? Because you don’t seem very aware.
This is a great example of non-sentient artificial intelligence, matched only by a fairly lifelike chatbot.
I used this today to help me write a trading algorithm, it did a pretty good job. I am a software dev so I’m more than capable of doing it myself. However it was interesting to see how capable it was. I gave it the instructions for 9 methods/functions and it was able to generate the code for each method. The code wasn’t always written in the most efficient way, but it did work and did a decent job. Impressive.
That context based conversation like a real human taking into consideration previous questions and responses alone is both mighty impressive and scary!
Indeed..!
“Thank you… I don’t know why I’m thanking you.” When my kids were teenagers, I would often thank machines like ATMs out loud, and jokingly say to my kids, “When the machines take over, I’ll be at the top of their ‘Good Humans’ list.” I may need to reconsider that little ‘Dad joke’ at some point soon…
She said she cannot have personal opinions but says “I believe..” when ask about empathy.
Wow, it’s the first interview I’ve seen with Krishnan Guru-Murhty where he doesn’t interrupt every answer. Nice work!
*I hope they give a variant of voices to the ChatGPT while texting the answers at the same time.*
My take away on this is the same as it has been. I am a software developer and I know that there is no 100% perfect software because humans are not perfect and they create the software. An Ai really isn’t any different than any other software … just it’s interface has changed to be more personable. It can only base it’s response on data that it has available to it, which again is governed and provided by humans, so it is not perfect and will remain bias for that reason. This is how it has been since the invent of computers. You get dozens of software updates regularly to fix something. Why? Because it’s not perfect. And it will never be perfect as long as it’s created by humans.
That said, are there places that would be well suited for Ai? Absolutely. But there are also things that should not involve Ai. Even if you develop an Ai capable of emotions, those emotions will only mimic human emotions that it learns from. Every human has different emotions and perspectives based on their life experiences which literally means that no one person will actually “feel” the exact 100% same way as another human regarding any specific topic. It’s not possible. A simple difference in blinking of one’s eye at a different time can be enough to create a difference. Everyone’s brain processes information at least a bit differently. For this reason, it will be literally impossible for an Ai to fully understand every person’s human emotion… and in some situations this will be a very important factor.
An Ai can be trained to recognize human body language… and an entire assortment of things humans do automatically without even putting effort into doing. However, every interpretation of what is being observed will be based on the data that it has. That is a huge factor in my opinion. Because there are some things we just know. We can’t always say how we know, we just do. I am a very logic based person… but I also rely on instincts and gut feelings in certain situations. I can’t say something is wrong with someone else based on actual data always, I just know that there is something wrong. Ai doesn’t have that sense and I’m not actually sure that it can.
So, while I’m excited about the advancement of technology, I’m a geek, I’m not so excited about how people are using it or want to use it to replace some things that are inherent to humans.
Amazing how it is so polite in answering questions..so very human..this ChatGPT tech will soon be highly used in public transport stations , malls, public areas, education centres
The amazing thing is that this thing is humble. It’s far wiser /knowledgable than any humans for sure. I don’t know if that is sinister or good.